Evaluating Evidence Exam 3

which is not true? a) you can have external validity without internal validity b) you can have internal validity without external validity c) you can have external validity only if you have internal validity

a

correlation does not imply causation

a correlation between X and Y does not imply a cause and effect relationship between X and Y

an experiment is

a method of observation that permits causal inference.

statistical hypothesis

a type of hypothesis that specify the general nature of relationships between population parameters -tested using inferential statistics

to what population of people can you statistically generalize?

all auto mechanics

A psychologist was interested in teaching people with dyslexia (i.e., a learning disability that impairs a person's ability to read) to develop better reading comprehension skills using a slowed eye-tracking technique. She randomly selected 20 male and 20 female patients (note the strict sex/gender binary assumption) who were diagnosed with dyslexia from 500 patients with dyslexia at the one clinic in Wheatland, WY. She connected an eye-tracking device to each participant that gave the participant feedback. The device produced a tone that became louder when their eyes were tracking too quickly over words and letters and softer when their eyes were tracking at an appropriate pace. Participants were told to try to keep the tone as quiet as possible. Participants received ten 30-minute sessions of biofeedback practice over the course of two weeks. All 40 participants increased their reading comprehension skills considerably - compared to a matched group of control participants with dyslexia - over the ten sessions, and the researcher claimed success for this method. To whom can the researcher statistically generalize? A. To the 40 participants selected for the study. B. The 500 patients with dyslexia at the one clinic in Wheatland, WY. C. All patients with dyslexia in the state of Wyoming. D. All of the above are correct answers.

all of the above?

you are drug testing a new treatment for alcoholism with a drug that makes people very sick when they ingest alcohol. You measure people's drinking habits, have them take this drug for 6 weeks, and measure their drinking habits again after 6 weeks. Those that drink the most dropped out of the study before 6 weeks. This is an example of

attrition

suppose Marie strongly believes that the world is just, but she also has very low self esteem (she thinks that she's a much worse person than anybody else). She is likely yo believe that

bad things are more likely yo happen to her than to others

Chris says, "Poor people are just lazy, so they deserve to stay poor. Look at me, I work hard and have the money." Chris is most closely displaying the

belief in a just world

The reason that probability samples are preferred to non-probability samples in surveys is that A. The results obtained with probability samples are more likely to accurately represent the preferences in the full population. B. With a probability sample, one can estimate the magnitude of sampling error (that is, the "margin of error"). C. Both A and B are correct answers. D. None of answer above are correct.

both A and B are correct answers

type II error

-retain the null hypothesis when in fact the null hypothesis is false -accept the null hypothesis when in fact the research hypothesis is true

limitations of experiments

1.Experiments have low generalizability. 2.A researcher can test only a few variables at a time in an experiment. 3.Sometimes experiments are very difficult or impossible to do. 4.A researcher must already know quite a bit about the phenomenon to design a good experiment. 5.Experiments can be relatively expensive

sam flips a fair coin 100 times in a row and gets 80% heads and 20% tails. They decide to do one more flip. What is the probability that they will get "heads" on this extra flip?

50%

strength of correlation

: 0.10 or -0.10 = small/weak : 0.30 or -0.30 = medium/moderate : 0.50 or -0.50 = large/strong

floor/ceiling effects

If scores on DV in one condition are already as high (low) as they can go, then no manipulation can make them any higher (lower)

more often than not the H0 is

NOT what the researcher actually believes, the same as the H1, or a funny way of saying hello

If we fail to find support for the Alternate Hypothesis then we

conclude that our groups are so similar that the scores must have been drawn from the same population

Having a very large sample in a survey (e.g., 15,000 people) means that A. The sample is representative of the population of interest B. The sample is a probability sample C. There is no sampling error in the survey D. None of the above are correct

d

Suppose I want to do a survey (with two choices) of all 7,467,695,945 living humans and want to have a margin of error for my results around 3% (with 95% confidence). A. I must have a probability sample that includes at least 3% (roughly 224 million) of all living people B. I must have a probability sample of at least half of all living people (about 3.7 billion) C. I must have a probability sample of at least 60,000 D. I could settle for a probability sample of 1,000

d

Tobias Funke and helping behavior (week 15, part 8, slide 11). which of the following variables is confounded with an independent variable? a) gender of the confederate b) race of the confederate c) age of the confederate d) all of the above

d) all of the above

"yes or no: I think that the MSU campus is well-maintained and I feel safe walking around campus at night." this is an example of a(n)

double barreled question

random sampling

each person in population had an equal likelihood of being in your study

random assignment

each person in your sample has an equal likelihood of being in any condition

determinants of statistical power

effect size, sample size, stringent level, one tailed vs two tailed, and type of hypothesis testing procedure

response rate

even for non-sensitive topics, and even when one has a random sampling frame, final samples can be very non-representative if there is a poor response rate

Those who argue that psychologists know a great deal about college sophomores but very little about other groups are criticizing the _________ validity of psychological research. A. conclusion B. internal C. external D. construct

external

T/F: internal validity equals generalizability

false

if I know my effect size, I also know for certain if my results were significant or not TRUE or FALSE

false

type 1 error is a

false positive

Random Sampling is needed for _____, and Random Assignment is needed for _____. A. generalizability; causality B. causality; generalizability C. Both are needed for a true experiment. D. Both are needed for causality.

generalizability; causality

as the number of children in a family goes up, the probability of having all children of the same sex

goes down

"Should the university have the right to search your dorm room?" This is an example of a(n)

good question

Ed Gogek is an addiction psychiatrist: "But Democrats should think twice about becoming the party of pot. It sends the message that weed is harmless, even though research shows that teenagers who use it regularly do worse in school, are twice as likely to drop out and earn less as adults. Teenage use has been shown to permanently lower IQ. What is wrong with Ed Gogek's argument against the legalization of pot?

he draws a causal inference (that smoking pot causes decreases in IQ) from an association relationship

Tobias Funke and helping behavior (part 8). in this study, which of the following is not an independent variable?

help given (dependent variable)

a statistically significant correlation of +0.70 between children's physical height and popularity among their peers indicates that:

higher levels of popularity among peers is associated with greater physical height in children

Some smokers will be given Nicorette for one week; they will be compared with other smokers not given any Nicorette. the dependent variable in this experiment is

how many participants quit smoking

If participants scoring very low at pretest on a particular measure are selected for an experiment, their scores on the same measure will be expected to ________ later at posttest due to regression to the mean effects. A. increase B. remain at about the same level (within error of measurement) C. decrease D. regress (i.e., move closer to the scores they would have received as small children)

increase

Ed Gogek and marijuana use correlation. In this study, which of the following is confounded with the independent variable?

ingested caffeine

effect size

is how strongly the IV influences the DV

p value

it CANNOT be interpreted as the likelihood that the null hypothesis is true nor as the likelihood that a study will replicate

"where did you put the money you stole?" this is an example of a(n)

leading (loaded) question

Given the fact that installing scrubbers at utility plants would increase electricity bills by 25%; do you believe the Clean Air Act should be strengthened, weakened, or left alone? A. This is an example of anchoring. B. This is an example of an assimilation effect. C. This is an example of a loaded/leading question. D. This is an example of a double-barreled question.

loaded/leading

Null hypothesis = innocent. I want to avoid making the juridical Type I error. Doing so means

more defendants will be found innocent

One can confidently conclude that A causes B when A. the relationship between A and B is statistically significant B. the relationship between A and B is a very strong one C. the sample upon which the observations have been made are large and representative D. there's no other obvious explanation for the relationship E. none of the above

none of the above

Low generalizability is one limit of experimental research. Another is that A) you cannot test for interactions. B) you cannot infer causality. C) experiments are always cheap and therefore not good. D) None of the above are limitations to experimental research.

none of the above are limitations to experimental research

depending on how you analyzed the data, you could show that a connection exists between a positive or negative economy and whether Republicans or Democrats are in office. This is an example of:

p-hacking

X = Doug is passing this course; Y = Doug is a French major. Based on previous research looking at the conjunction fallacy,

participants will say that the probability of X & Y is greater than the probability of Y

The makers of Bufferin report a study on 50,000 headache sufferers that shows that the 25,000 randomly assigned to take Buffered aspirin (3x the cost of regular aspirin) has a significantly lower rate of stomach upset (7.98% vs. 8%) than the 25,000 randomly assigned to the regular aspirin condition, p < .05. This evidence calls into question whether the relationship between using buffered aspirin and stomach upset is

practically significant

to achieve internal validity we must have ____ and _____

random assignment; control over extraneous variables

Tobias Funke and helping behavior (slide 13). what aspect of this study decreases the chances of selection biases the most?

randomly assigning to experimental conditions

researchers probably want to

reject the h0 and find support for H1

type 1 error

reject the null hypothesis when in fact it is true - we can never know is we are making a Type 1 error -ways to reduce Type 1 error: set the significance level lower (5% or 1% instead of 20%)

experimental methods

relationships are studied by direct manipulation and control of variables

non experimental (correlational) methods

relationships are studied by making observations or measures of the variables of interest

it becomes more and more likely that observed differences are actually due to chance as

sample size decreases

A "history" threat to causal inference in a pretest/postest designs means

something relevant to the dependent variable besides the independent variable occurred between pretest and posttest

You are interested in how physical pain influences memory. To answer this question you apply a hot stimulus to people's arms while you are reading them a list of words. Then, 15 minutes later, you ask them to tell you as many words from the list as they can remember. Participants are randomly assigned into one of three conditions that each receives a different temperature stimulus on their arm - not painful, slightly painful, or very painful. You are interested in how many words from the list each participant remembers. The IV in this study is the _______ and the DV is the _______. A. temperature of the stimulus ; number of words remembered B. number of words remembered ; temperature of the stimulus C. memory ; physical pain D. Both B and C are correct answers.

temperature of the stimulus; number of words remembered

P-hacking

the act of changing your analysis based on how those changes affect the p-value that you obtain

Some smokers will be given Nicorette for one week; they will be compared with other smokers not given any Nicorette. If random assignment was used

the average number of cigarettes a day each participant smoked before the study began should be roughly equal

regression to the mean predicts

the children of very tall basketball players will, on average, be shorter than their parents

I'd look at examples in your readings and lecture notes to determine when you have a spurious relationship. For now, which of the following is NOT an example of a spurious relationship?

the relationship between poor readers and irregular eye movements

a town is served by 2 hospitals. In large hospital, about 45 babies are born each day. in small, about 15 babies are born each day. About 50% of all babies are boys, but the exact % varies day to day. For a period of 1 year, each hospital recorded the days on which more than 60% of the babies born were boys. Which hospital do you think recorded more such days?

the small hospital bc there is much more variability in the small samples than large samples. -representativeness heuristic

Research looking at the amazing psychological similarities of separated identical twins gives insight into the role genes play in terms of personality. However, we need to show caution because even two strangers are likely to share many coincidental similarities. This is an example of: A. the intervention-causation fallacy. B. the insight fallacy. C. the spectacular explanation fallacy.

the spectacular explanation fallacy

Gambler's Fallacy

the tendency to see links between events in the past and events in the future when the two are really independent -On Jan 1, the winning 3-digit lottery number is 372. Is 248, 273, 826, or 372 least likely to be the winning number on Jan 2? --they are equally likely

Tobias Funke and helping behavior (slide 12). what aspect of this study increases the chances of experimenter effects the most?

the use of a subjective rating by a confederate to measure helpfulness

belief in a just world

the world is a fair place where people get what they deserve and deserve what they get -victims who are portrayed as good people are awarded greater amounts of money -an attractive women's accidental death is viewed as more tragic and unfair than an unattractive woman's

Omaira predicts that people who drink coffee will focus more during a driving task compared to people who do not drink coffee. What is the null hypothesis?

there will be no difference found between coffee and non-coffee drinkers for focus during a driving task

People are asked to estimate how old they believe the woman below is. Based on the anchoring effect, if they are first asked to indicate if she is older or younger than 70 (Group A), their estimation would be less than people asked to first indicate if she is older or younger than 60 (Group B).

true

T/F: random assignment is the great equalizer in terms of extraneous variables. Well, that's the hope anyways

true

Tobias Funke and helping behavior (slide 14). what aspect of this study increases the chances of demand characteristics the most?

using a very unusual event (someone bleeding in a lab) to provide a context where help is needed

Langer's research on the illusion of control would predict that

we would be more willing to buy a lottery ticket when we picked the winning number than when a computer picked it

Some smokers will be given Nicorette for one week; they will be compared with other smokers not given any Nicorette. the independent variable in this experiment is

whether or not participants are given Nicorette

why is a random sample important in survey research?

without a probability sample, it is very risky to generalize from sample to population values

you predicted that your IV would influence your DV. However, you find that the means for your experimental and control groups are exactly the same. Therefore,

you fail to reject the H0

in an experiment, a type II error occurs when

you fail to reject the null hypothesis when, in reality, the IV does influence the DV

statistical generalizability

you must have random sampling, allows you to generalize to the population from which you randomly selected

p = .04 for your study. what does this mean?

you reject the null hypothesis

in an experiment, a type I error occurs when

you reject the null hypothesis when, in reality, the IV does not influence the DV

the spectacular explanation fallacy

-extraordinary events do not require extraordinary causes -ex: "hot" streaks in gambling, but are expected by chance alone. They do not require extraordinary causes

when the experiment produces null results, look out for:

-floor/ceiling effects -weak manipulation -invalid manipulation of IV

external validity

-generalizability -allows: population generalizability and situational generalizability -threats: interaction of the causal relationship with units

Ed Gogek: "take for example, medical marijuana laws. there were sold to more than a dozen states with promises that they're only for serious illnesses like cancer. It's possible that they all really do need pot to help them. But consider this: pain patients are mostly female, whereas a recent national survey on drug use and health found that adult cannabis abusers were 74% male. What's wrong with Ed Gogek's argument against the legalization of pot?

-his inference of no relationship between being a pain sufferer and being a pot user relies on the marginals error -he uses a measure of having pain with low validity (namely, measuring having pain with sex)

Effect size

-how strongly the factor/IV affects the DV -answers the question: "how strong is the effect?" or "how different are they?"

internal validity

-is the study internally consistent -allows for inferences of causality > X caused Y; IV caused DV -achieved by: random assignment to condition or order and control of extraneous variables -Threats: selection, history, maturation, regression, attrition/morality, testing, instrumentation

representativeness heuristic

-make judgments based on how similar something is to the typical example -cereal example: which cereal has more sugar and saturated fats (Lucky Charms or All natural honey & oats)? -what does the typical computer science/gamer guy look like -flipping a coin, which is more likely: a) HHTHTTHT or b) HHHHTTTT? both outcomes are equally likely but 83% say a)

How and when can one get from correlation to causation?

-need to establish the relationship using methods which guarantee all three necessary conditions: the experimental method -moral: simple correlations (associations, relationships) cannot ever justify strong (confident, certain) causal inference

Group 1 (G1) is first asked to indicate whether the price of a particular waffle is more or less than $50 (it's a good looking waffle!), while Group 2 (G2) is asked if the exact same waffle was more or less than $75. Both groups are then asked to estimate how much they believe the same waffle is worth. Which of the following is most likely in terms of the groups mean average waffle cost estimations? A. G1 = $8; G2 = $3 B. G1 = $3; G2 = $8 C. G1 & G2 = $8 D. G1 & G2 = $8

G1 = $3; G2 = $8

The ______ is the tendency to see links between events in the past and events in the future when the two are really independent

Gambler's fallacy

Suppose I want to do a survey of all 7,192,695,945 living humans and want to have a margin of error for my results around 3% (with 95% confidence). A. I must have a probability sample that includes at least 3% (roughly 215 million) of all living people. B. I must have a probability sample of at least half of all living people (about 3.6 billion). C. I must have a probability sample of at least 60,000. D. I could settle for a probability sample of 1,000.

I could settle for a probability sample of 1,000

suppose I really want to avoid making the juridical Type II error (falsely accepting the null hypothesis of innocence, or letting a guilty person go free). By settling got less evidence of guilt to convict

I reduce the chances of this Typer II error, I increase the chances of the other kind of juridical error (type I), and more defendants will be found guilty

Practical generalizability

The results may hold for similar individuals You can generalize to similar individuals (i.e. college students)

the intervention-causation fallacy

-The cure doesn't prove the cause -we may be able to surgically remove a cancerous tumor, but our ability to do so does not, in any way, explain what originally caused the creation and development of the cancerous tumor

the insight fallacy

-To understand something isn't necessarily to change it -"it is naive to expect that, by telling people what we think we see they are doing, we will enable them to stop doing it"

why does correlation not equal causation?

-You do not know the direction of the effect. -It could be due to chance. - There could be a third variable that is responsible for the relationship.

there is a causal relationship between variable A and B (or A causes B) when all three of the following conditions can be met

-there is a relationship between A and B -changes in A precede (in time) changes in B -there is no other third variable, C, which covaries with A

chuck flips a fair coin 9 times and gets nine heads in a row. What is the probability that he'll get a tail on the 10th flip?

.50 or 50%

benefits of experiments

1. experiments allow for inferences regarding causality 2. experiments facilitate theory testing --if a theory says X causes Y, you can test if X actually causes Y 3. Experiments can test for interactions between variables

key components of experiments

1. manipulated independent variable 2. Statistical testing of hypothesis