1. Introduction To Hypothesis Testing
Chapter 5
Inferential Statistics: Making inferences about populations based on samples

2. Chapter Outline Hypothesis-Testing
The Core Logic of Hypothesis Testing
The Hypothesis-Testing Process
One-Tailed and Two-Tailed Hypothesis Tests
Decision Errors
Hypothesis Tests as Reported in Research Articles

3. Hypothesis Testing
A systematic procedure for deciding whether the results of a research study (using a sample) supports a hypothesis that applies to a population (probabilistic conclusion)
Hypothesis: A predictions tested in a research study based on informal observation or theory
e.g., Concrete words are remembered better than Abstract
Theory: a set of principles that attempts to explain one or more facts, relationships, or events usually gives rise to various specific hypotheses that can be tested in research studies
e.g., Dual-coding Theory: Concrete 2 codes & Abstract 1 code

4. Hypothesis Testing Researchers want to draw conclusions about a
particular population
e.g., Babies in general (POPULATION)
Conclusions will be based on results of studying a
sample
e.g., One baby (SAMPLE)

5. The Core Logic of Hypothesis Testing
Researchers must spell out predictions in advance (A priori) to conclude that their hypothesis was supported.
Establish the p-value (a priori) of getting our experimental results if our hypothesis was not true.
Researchers attempt to answer these questions:
What is the probability of getting our research results if the opposite of what we predicted were true?
If it is highly unlikely that we would get our research results if the opposite of what we are predicting were true:
We can reject the opposite prediction.
If we reject the opposite prediction, we can accept our prediction.
Then we conduct the experiment.

6. Hypothesis Testing Principle
Are there differences between men and women in regards to IQ tests? Make a prediction by answering question!
Research Hypothesis or Alternative Hypothesis (HA)
IQM ≠ IQW or IQM < IQW or IQ M > IQ W
HA: Predicted Relationship between the populations
There is a difference: IV affecting the DV!
Null Hypothesis (H0): IQ Men = IQ Woman
H0: Relationship between populations that is the opposite
of the Research Hypothesis (Scientific Hypothesis)
It simply says: No change, nothing happens, hence NULL

7. Hypothesis Testing Rejecting vs. Fail to Reject the Null Hypothesis
Always test the null hypothesis: Test the opposite (i.e., no difference) of what you predict
Fail to Reject (Accept) H0: No difference, inconclusive
Reject H0: There is a difference, HA is supported
You want to always REJECT H0

8. Example
A person says that she can identify people of above average intelligence with her eyes closed. Is this true?
STEP1. Reframe question into a research and null hypothesis about populations.
Population 1: People chosen by woman with her eyes closed (THIS IS THE SAMPLE)
Population 2: People in general (Comparison Distribution, KNOWN)

9. Research Hypothesis (HA): Those chosen are more intelligent
Population 1 > Population 2
Population 1 has a higher mean intelligence than population 2
Null Hypothesis (H0): Those chosen are not more intelligent
Population 1 = Population 2
Population 1 does not have higher mean intelligence than population 2.

10. STEP 2. Determine Characteristics of the comparison distribution
Known normal distribution with Pop. M = 100; Pop. SD = 16
STEP 3. Determine the Cutoff Sample scores on the comparison distribution (Criterion has to be done A PRIORI) at which the null hypothesis should be rejected (p < .05)
5% probability (top 5%), Z needed is 1.64
50% -5% = 45% >1.64

11. Null Hypothesis Significance Testing (NHST)
p = .05 indicates that, assuming the null hypothesis is true there is a 5% probability of obtaining a study score as extreme, or more extreme, than our study prediction. This indicates our study prediction is unlikely when the starting assumption (the null hypothesis) is true. If the p-value is low enough (< .05) we reject our starting assumption – the null hypothesis.

13. STEP 4. Determine the score of your sample on the comparison distribution
Person picks out individual with IQ score = 140
Z= (x – Pop. M)/Pop. SD = ( )/16 = 40/16 = 2.5
5. Compare the scores obtained in Step 3 and 4 to decide whether to reject the null hypothesis
Score on 4 (Z = 2.5) is higher than score on Step 3 (Z = 1.64)
Conclusion: Reject H0: HA is supported

17. Implications of Rejecting or Failing to Reject the Null Hypothesis
When you reject the null hypothesis all you are saying is that your results support the research hypothesis.
The results never prove the research hypothesis or show that your hypothesis is true.
Research studies and their results are based on the probability or chance of getting your result if the null hypothesis were true.
When you fail to reject or accept the null hypothesis, you do not say that the results support the null hypothesis.
You say that the results are not statistically significant, or that the results are inconclusive.
We are basing research on probabilities, and the fact that we did not find a result in this study does not mean that the null hypothesis is true.

18. Significance Level Criterion as % cutoff or p value
The lower the significance level, the more sure we are in rejecting H0: Top .01% (p < .0001): Z score = +3.5)
The higher the significance level, less likely to deal with inconclusive results: Top 50% ( p = .5, Z = 0)
and always reject the H0:

20. Significance Level Convention in Psych: p < .01 or .05
How results are expressed:
Reject Ho: Results are statistically significant
Fail to reject Ho: Results are not statistically significant

21. One- and Two-Tailed Tests
Researcher predicts direction of HA < or >
ANACIN improves memory
Experimental Group will score higher than Control group
Population 1 = group received Anacin
Population 2 = group did not receive it
HA= Pop. 1 Remember more words than Pop. 2
Ho=Pop. 2 not remember more words than Pop. 1

22. One- and Two-Tailed Tests
One-Tailed Test: A directional hypothesis: it predicts a direction of result.
Also called a one-tailed test: hypothesis test looks for an extreme result at just one end (high or low end) of the curve.
Psychic (high end)
Stres-reduction-method example (low end)

26. Two-Tailed Tests Experimental treatment will make a difference:
Two-tailed test (non directional): Doesn't sayhow high or how low.
Extreme results: Either extreme or tail (a lot or a little sleep) would make us reject the Ho
Example: A polluting substance accidentally put in the water in a particular region may be suspected to affect the part of the brain that controls sleep, but it is not known whether it will increase or decrease amount of sleep.

30. Non-directional hypothesis: polluting substance to affect sleep BUT no direction of effect specified.
In setting the cutoff for rejecting H0
p < .05) value divided between the two tails
.05/2 = .025 or 2.5% at each side.
To be significant, score more extreme than if a one-tailed test

31. One-Tailed or Two-Tailed Tests
Use a one-tailed test when you have a clearly directional hypothesis (Theory Driven)
Use a two-tailed test when you have a clearly non-directional hypothesis (Don't know which way).
With a one-tailed test, if the sample score is extreme—but in the opposite direction—the null cannot be rejected.
Often researchers will use two-tailed tests even if the hypothesis is directional (As protection!)

32. Decision Errors When the right procedures lead to the wrong decisions
In spite of calculating everything correctly, conclusions drawn from hypothesis testing can still be incorrect.
This is possible because you are making decisions about populations based on information in samples.
Hypothesis testing is based on probability.

33. Hypothesis Testing Principle
A person says that she can identify people of above average intelligence with her eyes closed. Is this true?
1. Population 1: People chosen by woman
Population 2: People in general (Known)

34. Research Hypothesis (HA):
Population 1 > Population 2
Null Hypothesis (H0):
Population 1 = Population 2
2. Characteristics of the comparison distribution
Known normal distribution with µ = 100; σ = 16

35. TYPE I ERROR α
3. Determine the Cutoff Sample scores on the comparison distribution (criterion) at which the null hypothesis should be rejected (p = .5)
50% probability (top 50%), Z needed is 0
A 30% Probability (top 30%)? Z = .524

36. Type I (α) Error
Rejecting the null hypothesis when the null hypothesis is true
You find an effect when in fact there is no effect.
A Type I error is a serious error as theories, research programs, treatment programs, and social programs are often based on conclusions of research studies.
The chance of making a Type I error is the same as the significance level.
If the significance level was set at p < .01, there is less than a 1% chance that you could have gotten your result if the null hypothesis was true.
Reduce chance of making a Type I error by decreasing significance level (e.g., p < .001).

37. Type II Error
With a very extreme significance level, there is a greater probability that you will not reject the null hypothesis when the research hypothesis is actually true.
concluding that there is no effect when there is actually an effect
The probability of making a Type II error can be reduced by setting a very lenient significance level (e.g., p < .10).

38. Outcomes of Hypothesis Testing
Type I error
Reject the null hypothesis when it is correct.
Saying there is a difference when there is not a difference
We are rejecting a true null hypothesis.
Accepting a research hypothesis when in fact it is not true

39. Acceptance
Area
Rejection
Area

40. Acceptance
Area
Rejection
Area

41. Relationship Between Type I and Type II Errors
Decreasing the probability of a Type I error increases the probability of a Type II error.
The compromise is to use standard significance levels of p < .05 and p < .01.

42. Type II Error ( ß).
Not rejecting HO when it is false.
Accepting HO when it is false.
Saying there is not a difference when in fact there is a difference
Why?
Because the critical value (p < .001) may be too extreme or low, and maybe sample representative.
There maybe a difference but p value to low.

43. Probability of making a Type I error (α)?
The p = value chosen (p < .05):
Probability of Type I error is 5 out of 100.
This is why we do not use p’s = .2 (20 out 100).
The smaller the prob of Type I Error the more we are controlling: p < .001
A p = .2 increase Type I error.

44. TYPE II ERROR β
3. Determine the Cutoff Sample scores on the comparison distribution (criterion) at which the null hypothesis should be rejected (p = .05)
5 % probability (top 5%), Z = 1.64
2.5 % Probability (top 2.3 %), Z = 1.96
1 % Probability (top 1%), Z = 2.33
How about a p < .001
or a p < .0001

45. Concluding that there is no effect when there is actually an effect
Type II (ß) Error
With a very extreme significance level: Greater probability to not reject H0 when the Ha is actually true.
Concluding that there is no effect when there is actually an effect
Reduce Type II error by increasing p- value (e.g., p < .10)

47. Acceptance
Area
Rejection
Area

48. Acceptance
Area
Rejection
Area

49. Acceptance
Area
Rejection
Area

50. Acceptance
Area
Rejection
Area

51. Acceptance
Area
Rejection
Area

52. Acceptance
Area
Rejection
Area

53. Rejection
Area
Acceptance
Area

54. Type II Error a Very High Price to Pay for
Theory building:
Sometimes we may abandon a particular good theory because of a small p value.
Practical applications of a particular treatment
Which one to use?
Control Type I error by using small p (.0001)
Control Type II error by increasing p (.2 or .3)
Use conventional values: .01 and .05.

57. Relationship Between α and β Errors
Decreasing the probability of a Type I error increases the probability of a Type II error.
The compromise is to use standard significance levels of p < .05 and p < .01.

60. When to Use One-Tailed or Two-Tailed Tests
Use a one-tailed test when you have a clearly directional hypothesis.
Use a two-tailed test when you have a clearly non- directional hypothesis.
With a one-tailed test, if the sample score is extreme—but in the opposite direction—the null cannot be rejected.
Often researchers will use two-tailed tests even if the hypothesis is directional.
Copyright © 2011 by Pearson Education, Inc. All rights reserved

61. Hypothesis Tests as Reported in Research Articles
In research articles, for each result of interest, the researcher usually says whether the result was statistically significant.
The researcher gives the symbol for the specific method used in figuring out the probabilities.
There will be an indication of significance level (e.g., p < .05 or p < .01).
Usually a two-tailed test is used; if this is not the case, the researcher will generally specify that a one-tailed test was used.
Copyright © 2011 by Pearson Education, Inc. All rights reserved

62. Copyright © 2011 by Pearson Education, Inc. All rights reserved
Key Points
Hypothesis testing considers the probability that the results of a study could have come about even if the experimental procedure had no effect. If this probability is low, the scenario of no effect is rejected and the hypothesis behind the experimental procedure is supported.
The expectation of an effect is the research hypothesis; the hypothetical situation of no effect is the null hypothesis.
When a result is so extreme that it would be very unlikely if the null hypothesis were true, the null hypothesis is rejected and the research hypothesis is supported. If the result is not that extreme, the researcher does not reject the null hypothesis and the study is inconclusive.
Behavioral and social scientists usually consider a result extreme enough if it is less likely than 5% that you would get that result if the null hypothesis was true. A significance level of 1% is also commonly used in research.
The cutoff percentage is the probability of the result being extreme in a predicted direction in a directional or one- tailed test. The cutoff percentages are the probability of the result being extreme in either direction in a nondirectional or two-tailed test.
Steps for hypothesis testing are:
Restate the question as a research hypothesis and a null hypothesis about the population.
Determine the characteristics of the comparison distribution.
Determine the cutoff sample score on the comparison distribution at which the null hypothesis should be rejected.
Determine your sample’s score on the comparison distribution.
Decide whether to reject the null hypothesis.
A Type I error is when a researcher rejects the null hypothesis but the null hypothesis is actually true. A Type II error is when a researcher does not reject the null hypothesis, but the null hypothesis is actually false.
Research articles report the results of hypothesis testing by saying whether the results were significant, giving the cutoff sample score on the comparison distribution, and giving the probability level of the cutoff sample score.
Copyright © 2011 by Pearson Education, Inc. All rights reserved