Final Exam Review
∙ Learn how to make questions concrete (turning big questions in small questions)
-Know what the terms big and small questions mean and their relationship together
-We put them in standard form (what is the effect of?) ∙ Know what independent and dependent variable mean
∙ Know the different types of research (ex: case studies, archival studies)
∙ Naturalistic Observation
∙ Experimental research control (well controlled: get rid of confounds); absence of something (confounds); can control what happens over the study
∙ Fundamental threat for good research confound variable -Confounding Variable: systematically covered with IV and tell plausible story why this impacts DV
Don't forget about the age old question of What is true about metabolic pathways?
-How to avoid compounds: randomize participants to groups, so IV is manipulated into experiments between the groups (one gets experimental condition while the other gets controlled; between subject design)
∙ Factors: IV and various conditions of levels
∙ Significance: levels of various conditions differ in terms of their outcomes; dependent scores differ from experimental or control conditions
∙ Repeated Measure design Don't forget about the age old question of Why did jefferson and hamilton hate each other?
∙ Internal validity (well-designed study with no compounds) ∙ External validity (Can generalize study)
∙ Measures of Central Tendency (tells you what most people are doing or what data is doing; measure that represents everybody)
-Mean (measure that is used the most; average of group) -The arithmetic average
-The score above which 50% of the scores lie and below which 50% lie
-If you put numbers in a row, it would be the one in the middle one; if there are two of the same numbers then it would be the even number
-In a skewed distribution is when we would want to use the median most
-The most common score (frequently reported value)
∙ Describing Data
-Range (how big is the difference from the lowest value to the highest) We also discuss several other topics like What is the meaning of “total war”?
-The Variance (measure of how spread out the data is; subtract each score by the average score and then square that value)
-The Standard Deviation (square root of the variance)
∙ Outcomes: Interactions
-Suppose a drug works well for older women but not for younger women
-Support that drug works well for younger men but not for older men
-Do we give the drug to women? Men?
-Do we give it to young people? Old?
-We give it to older women and younger men
-So… In a three factor study, there are SEVEN factors to evaluate Don't forget about the age old question of What are the economic variables of the development of russia?
∙ Frequency Distribution
-Normally in a graph the Dependent Variable is plotted on the graph
-The value of the range depends upon just two scores (highest and lowest)
-We would like a measure of dispersion that uses all of the scores in the distribution.
∙ Dispersion Average Deviations
-Columns of deviations squared
∙ A Measure of Dispersion/The Variance
-n=variance (number of scores in the distribution) -variance measures how spread out it is (the bigger the variance, the more spread out it is)
-sum of the variation of the means
-average of the squared deviation from the means - X with underline on top is the mean of the distribution
∙ The Standard Deviation
-Physicist and financers use this deviation; it’s a very common equation
-It is the square root of variance
-Represented by a little sigma symbol (o with curve; little s) -Variation of 10 = 3.16
∙ The statement on pg. 246 that states “the standard deviation…indicates the average deviation of scores from the mean” is NOT true. The Variance measures that.
∙ The “Normal” Curve
-frequency distribution Don't forget about the age old question of What model shows the flow of memory between 3 different types of memory?
∙ Population and Sample
-Key Point: we don’t really care so much about the sample (and therefore the statistic)
-We want to generalize from the sample to the population. That leads to the possibility of error.
∙ Decision with Uncertainty
-Your alpha (a) level
-If you set it at .05, that means that, on average, you have a 5% chance of concluding that the clinical has the skill when he is just guessing-that is you have a 5% chance of making this error (saying that the clinician has a skill, when he does not)
-If you set alpha at .01, then you have a 1% chance of making this error.
∙ In Other Words…
-If a result would be very rare (less than a) if it just happens by chance, AND
-if that result actually happens, THEN
-we say it was not just the operation of chance INSTEAD -we drop our skepticism and accept the claim
∙ Summary of the Decision Logic (advances of rational thought but not guaranteed to be right; you have to know you are running this at risk)
-A. There is a claim
-Research hypothesis (High > Low) If you want to learn more check out What is the albedo in climate?
-B. You doubt the claim (skeptic)
-You deny that there is a difference in performance between High and Low
-In other words, you say, Hi=Low. This is the Null Hypothesis
-C You set your alpha level (decision rule)
-D. You collect the data
-E. Compute how likely the results could have come out that way due to chance alone
-F. Decide, following your decision rule, whether to reject the null hypothesis
-We reject the null hypothesis when: the outcome is rare (less than a) if only chance is at work. If the result is rare enough, we say it was not just due to chance.
∙ Testing the Null Hypothesis
Consider the Null Hypothesis: the two groups are equal. Or, what is the same thing, the difference between them is zero (null).
-Draw a sample and compute that difference.
-It is unlikely to be exactly zero.
-How big does it have to be before you reject the null hypothesis?
∙ Characteristics of the Sampling Distribution -The mean of the Sampling Distribution is equal to the true mean (u) of the population. (Has been proved)
-To say it again, the mean of the sampling distribution, that is, the mean of the distribution of sample means = u -Therefore, if the Null Hypothesis is correct, the mean of the sampling distribution = 25.
-Also, the standard deviation of the sampling distribution can be estimated very accurately from the actual data that we gather.
-The standard deviation of the sampling distribution is called the Standard Error, SE (which is the sampling distribution). -It is approx. equal to the standard deviation of the actual sample divided by the square root of N, the sample size. -SE is the standard deviation of an almost imaginary distribution, the sampling distribution, but we can compute what it is!
-Again, the standard deviation of the sampling distribution (distribution of sample means), called the Standard Error, SE, is (approximately) equal to the Standard Deviation of one sample we collected, divided by the square root of N, the sample size. -We can use what we know about the normal distribution now to make inferences.
-Consider the Sampling Distribution
-It is normal
-If the null hypothesis is correct, its mean is 25.
-It’s standard deviation, SE- 1.5
∙ Testing the Null Hypothesis
-The mean of our sample = 28
-How many “units” is 28 away from the mean, if the null hypothesis is true?
-That is, what is the Z-score of 28?
-If the null hypothesis is true, is the score of 28 common or rare?
-How often will we see a score 2 units above the mean? -About 2% of the time. That is rare; it is less than “a.” -Therefore, we REJECT the hypothesis that our sample came from that distribution.
∙ Testing the Null Hypothesis: Summary
-Random sample from population; compute mean and SD -Compute unit size (SE) for the Sampling Distribution -Determine how many units the sample mean is from the mean of the sampling distribution, “u”, if the null hypothesis is correct. In other words, compute the Z-score associated with the sample mean.
-Determine what percentile this corresponds to.
-Decide if that percentile is smaller than “a.”
∙ Descriptive z-score
Z= x- x with underline on top/ “o”
∙ Inferential z-score (Mean if the null is true (Mo)) Z= x with underline on top – Mo/SE
∙ Maximize benefits, and minimize possible harmful effects to participants.
-Maximize for whom? Participants?
-What type of harm? Psychological/Stress or physical? Loss of privacy
∙ Autonomy: Individuals can decide for themselves; principle of informed consent
-Not necessarily informed about everything. Deception okay?
∙ Informed Consent
-Purpose of the research
-Risks and benefits
-Assurance of voluntary participation
-Is it fair and just for these participants to bear the burden of participating.
∙ Institutional Review Boards (IRB)
-Every institution in the U.S. that takes federal money, and that does research on animals or humans, are required to have an institution review board.
-Whenever anyone wants to do research, they have to describe in great detail and submit it to this organization. -Includes member of clergy, people who are sensitive when it comes to ethics, members of the community…these people are trained to say what is appropriate and aloud to do in the study and what is not aloud.
-This helps but it doesn’t completely rule out bad things
∙ Judgement Under Uncertainty: Heuristics and Biases
-What is the likelihood that object or event A
belongs to class B or process B?
-If A is representative of B, then we think that the likelihood is high
-Consider words of three letters or more. Are there more such words that begin with “r” or more than have an “r” as the 3rd letter?