Unformatted text preview: Basic Principle Probability, Samples & Populations Examine the probability that the outcome of a study might be possible even if the true situation was that the experimental manipulation made no different (i.e., had no effect). Probability
Cannot determine definitive truth or falsity of theories.
Never use the "p" word... word... Probability
The likelihood of an event occurring
Examples coin flip, pick a number... number... Therefore, we need to use probability to determine the likelihood of an event occurring. Probability
Range:
Proportion: 0 to 1.00 Percentages: 0%  100% Probabilities and Sampling Distributions
All statistical tests rely on sampling distributions to determine the probability that results are consistent with the null hypothesis Probabilities as symbols:
p p < 0.05 1 Examples of different sampling distributions Hypothesis Testing
How do we know if we have a true effect or just something due to chance? Does the outcome of our study support our theory? What is a "true" difference? What is a "true" difference?
In psychology, by convention we tend to say... say...
As long as there is less than a 5% chance that we made an error then we're confident with our finding. we'
Also denoted as p < 0.05 Difference! Null is true Difference! Probability and Sampling Distributions
When it is highly unlikely that the null hypothesis is true (usually <0.05 probability) then we reject the null hypothesis.
Still a chance we made an error
Type I Error / Type II Errors (coming up...) up... Probability
Cannot determine definitive truth or falsity of theories. Can use inferential statistics to make probabilistic conclusions. How do we work with this?
Replication Effect Sizes 2 Turning the page...
1. 2. What Type of Test? Depends on # of IVs & Measurement Scale Describing data Statistics used to make inferences about a population One IV 2 Groups One IV > 2 groups > Two IV Nominal Data Ratio/Interval Data Nominal Data Interval/Ratio Data Nominal Data Interval/Ratio Data ChiSquare Test of Association Independent ttest ChiSquare Test of Association Oneway ANOVA ChiSquare Test of Association Twoway ANOVA Dependent ttest Oneway ANOVA What's wrong with descriptive statistics? Consider the following... 120 babies/day 11 babies/day : What's wrong with descriptive statistics? Populations & Samples Sample Population M = 15.90 SD = 2.54 M = 20.1 SD = 5.94 3 Populations & Samples: Parameters Moving from Samples to Populations: Why be careful?
Representative Sample: Trying to generalize your findings back to the population
External Validity: ability to generalize
KEY = Random Sampling Be Careful! Moving from Samples to Populations: Why be careful?
Representative Sample: Trying to generalize your findings back to the population
External Validity: ability to generalize
KEY = Random Sampling
1. 2. Moving from Samples to Populations: Why be careful?
Internal Validity: Ability to make causal connections
Manipulated IV Random assignment Sampling Error:
What's due to chance (sample error) + what's due to What' what' your effect (effect of IV) Moving from Samples to Populations: Why be careful?
Internal Validity: Ability to make causal connections
1. 2. 3. Moving from Samples to Populations: Why be careful?
Internal Validity: Ability to make causal connections
1. 2. 3. Manipulated IV Random assignment Control ALL other variables
Confound: all other variables; "noise" in your data noise" Manipulated IV Random assignment Control ALL other variables
Confound: all other variables; "noise" in your data noise" 4 Does a true experiment exist?
For both internal and external validity, not "all or none" but rather both are continuous in none" nature Hypothesis Testing
Method for deciding whether effects are due to chance or to a true effect of the IV (or relationship between DV if correlation study)
If the probability of our theory being related to chance is low, we reject the null hypothesis KEY examine the probability that the outcome of a study could have occurred EVEN IF the "true" situation was that true" the experimental treatment made no difference. Basic Principle
Examine the probability that the outcome of a study might be possible even if the true situation was that the experimental manipulation made no different (i.e., had no effect). 5 ...
View
Full
Document
This note was uploaded on 04/07/2008 for the course PSY 031 taught by Professor Dicorcia during the Spring '08 term at Tufts.
 Spring '08
 Dicorcia

Click to edit the document details