Inferential statistics: - PowerPoint PPT Presentation

1 / 10
About This Presentation
Title:

Inferential statistics:

Description:

Inferential statistics: 1. Techniques for making generalizations from samples. Hypothesis testing: 1. Determining if the sample is representative enough of the ... – PowerPoint PPT presentation

Number of Views:147
Avg rating:3.0/5.0
Slides: 11
Provided by: COSS2
Category:

less

Transcript and Presenter's Notes

Title: Inferential statistics:


1
Inferential statistics
  • 1. Techniques for making generalizations from
    samples.

2
Hypothesis testing
  • 1. Determining if the sample is representative
    enough of the population so that an inference can
    be made.

3
5-step process for refuting chance (sampling
error)
  • 1st. Identify the characteristics of data
    collected, I.e., variables

4
2nd State your hypothesis
  • (a) research or alternate hypothesis, i.e., the
    expected relationship
  • (b) null hypothesis, i.e., statement of no
    relationship, no association, or independence

5
3rd Select sampling distribution
  • Use a Z distribution, whenever N is greater than
    120
  • Use a t distribution whenever N is 120 or less
  • 3. Use a one-tail testif the direction of the
    difference can be predicted
  • 4. Use a two-tailed testif direction cannot be
    predicted

6
4th calculate test statistic
  • Calculate the outcome for your sample, e.g.,
    regression, chi-square etc.

7
Level of Significance
  • The probability of drawing by chance an unlikely
    sample outcome. You can set the alpha at any
    level, but the convention is to set at .01 or .05
  • 2. Alpha .01 means 1 out of 100
    samples(times) you would be wrong to conclude
    that the sample differs from the population
  • 3. Alpha .05, i.e., 5 out 100 samples, etc

8
Critical region
  • 1. Rejection region. Once alpha level is
    selected, you can determine the sample outcome
    that begins the critical region by using the
    table--

9
5th make decision on null and research hypotheses
  1. Decision on Ho
  2. Reject a null hypothesismeans chance of
    an incorrect decision, I.e., type 1 erroris
    equal to the alpha level you have selected
  3. accept (fail to reject) means chance of
    type 2 error

10
Decision on research on research hypothesis
  • 1. Ha is there a statistically significant
    difference?
  • yes no
  • reject
  • accept
  • If you accept then you can conclude that there is
    a statistically significant difference
  • 2. Note you would either conclude that the
    sample is or is not representative of the
    population from which it was drawn. Implications--
Write a Comment
User Comments (0)
About PowerShow.com