PSYC 220 FINAL Flashcards
Numerical order for 1st row of counterbalancing
1, 2, n, 3, n-1, 4, n-2, 5, etc.
How to calculate for the largest number of people that can be used?
Divide number of subjects/levels into number of participants given in the problem and then multiply the answer by the number of subjects
Quasi-experiments
do not have to assign participants to levels, they already come in with those levels
How to write null hypothesis
“The mean”….”DV”……“participants”….”levels of experiment”…“WILL BE EQUAL.”
How to write out alternate hypothesis
“The mean”….”DV”……“participants”….”levels of experiment”…“WILL NOT BE EQUAL.”
How to write out interaction statement after writing down null/alternate hypothesis
The interaction between Level1 and Level2 will/will not be equal.
Analyze data from multi-factor, BS, ANOVA IF the interaction were significant
The interaction between Level1 and Level2 was significant, F(df1,df2) = f-value, p=exact p-value or if .000 write p<.001, ETA SQUARED
(We reject the null and accept the alternate hypothesis)
A simple effects test is necessary to determine where the significant differences lie.
How to calculate eta squared
Number of interaction total/number of corrected total
REMEMBER FOR ALL MATH, 2 DECIMAL PLACES
Analyze data from multi-factor, BS, ANOVA IF the interaction were NOT significant
The interaction between Level1 and Level2 was not significant, F(df1,df2) = f-value, p=exact p-value or if .000 write p<.001, ETA SQUARED
(We retain the null hypothesis)
Go ahead and analyze the poc hoc for each of your levels.
6 keys things in creating a graph
- Legend
- Footnotes
- Abscissa
- Ordinate
- Caption that describes contents of the graph, error bars, and sample size
- Origin value is zero
For bar graphs
same rules apply, but also:
- used when levels of IV are categorical, not numeric
2 used for single-factor and multi-factor designs
For single-factor, between subjects SPSS analysis significant
- “The difference among mean”…“DV” ….. was significant, F(df1, df2) = ans, p=exact p=value, if .000, then p<.001, eta squared
- (WE REJECT THE NULL AND ACCEPT THE ALTERNATE HYPOTHESIS)
- Analyze post-hoc test.
- The difference among mean….was significant, p=…..
- The difference among mean…was not significant, p=….
For single-factor, between subject SPSS analysis not significant
- “The difference among mean”…“DV” ….. was not significant, F(df1, df2) = ans, p=exact p=value, if .000, then p<.001, eta squared
- (WE RETAIN THE NULL HYPOTHESIS)
How to calculate expected frequency for chi square in a one-sample case
total number of subjects (N)/by the number of categories (k)
How to calculate expected frequency for chi square in a two-sample case
(rows total)x(column= total)/(total observations)
Df for one-sample case
n-1
Df for two sample case
(rows-1)x(columns-1)
example: 2 rows 2 columns (2-1)x(2-1)= (1)x(1)= =1
Analysis of chi square if greater than critical value found in the table:
Example:
The ……..were significantly different χ²=(df,N=total observations)=computed chi square value, p<.05.
(The computed χ² is greater than the critical value, (write value in parenthesis), at (whatever level, such as .05 level or .01 level)
Analysis of chi square if less than critical value found in the table:
Example:
The ……..were not significantly different χ²=(df,N=total observations)=computed chi square value, p>.05.
(The computed χ² is less than the critical value, (write value in parenthesis), at (whatever level, such as .05 level or .01 level)
For correlation statements
Low weak little
Medium moderate some
High High Strong
Last statement of correlation analysis
The correlation between A and B is/is not significant r(df)=correlation value, p=exact p value or if .000, report as p<.001
What is the only method that can determine causation?
Experiment
T-tests and ANOVAs will have at least how many per group?
30 (for really great power)
Correlations tests require at least how many pairs of measurements to be meaningful
10
What are examples of order effects?
practice, warm up, learning, boredom, fatigue, experimenter effects, or demand characteristics
How can you control order effects?
Through counterbalancing
Finding a difference in a sample when there is no difference in the population is called?
Alpha
the probability of correctly rejecting the null hypothesis is called?
Power
This is affected by random error only
Reliability
Failure to find a difference in a sample when there is a difference in the population is called?
Type II error
A measurement’s degree of accuracy is called?
Validity
What are the four basic goals of psychology?
Describe (What?, observations)
Understand (Why, translates into determining the cause)
Predict
Control
Understanding is limited to which type of research?
Experiment
T or F: The only difference between experimental psychology and other sciences (e.g. Chemistry, Biology, Physics) is the subject matter under investigation.
T, same procedure and logical processes used to reach goals
T or F: One of the reasons why people question whether psychology is scientific is because measuring the phenomena we are interested in is so easy that it doesn’t seem scientific
F
T or F: Another reason why people question whether psychology is a science is because we already have first-hand experience of many of the topics that are studied
T
T or F: Correlational and quasi-experiments are never demonstrations of cause
T
The better that we can predict a given outcome, the better the?
CONTROL
Reasons why people think psychology is not a science?
- See psychologists in applied settings
- difficulty of measuring (operational definitions)
The —-is used to describe the middle score in a set of scores.
Median
Your score is in the 34th percentile indicating that 34% of the scores in the population are
equal to or below your score
A score that is below the mean will have a z-score that is
negative
T or F: Because it is expressed in the same unit of measurement as the variable, variance is the easiest statistic of variability (or spread) to interpret
F
All measurements contain the true value of the object plus or minus whatever …exists.
error
The….of a measurement is defined as the ability of a measuring device, test, or procedure to produce consistent measurements when measuring the same event
reliability
Validity is affected by
both systematic and random error
How many SDs a given score is above or below the mean is what?
z-score
Percentiles, which are the percentage of the scores in the population that fall at or below, is derived from what?
z-score
Larger samples are better estimates of what?
population
They improve the accuracy of means, SD, and z-scores
All measurement has the potential for what?
error
Systematic error?
repeated error on every measurement
Random error?
unpredictable, changing, unknowable, estimated by SD
Only with…can we determine the degree to which we agree or disagree with other people
data
Nonparametric tests can be used for which two scales of measurement?
nominal and ordinal
Data?
available to the senses, publicly available, we could be able to see it
Non-data
subjective events, such as deductions, conclusions, and opinions are not data, beliefs, hypotheses, theories, adjectives
Example: Is this data or non-data?
Sara us very shy when she is around her friends
Non-data
Tommy said that the food was yummy
Data
She drove her car every day last week
Data
That perfume smelled really nice!
Non-data
Nominal
each is placed in a category from 0-1, no distinctions among group members
Ordinal
rankings, can know from most to least, but not how much, distinction can be made among group members
Interval
DO NOT START AT 0 1. can have distinction 2. can tell how much more or less 3. parametric test 4.
Ratio
SCALE STARTS AT 0
- YES, THERE IS DISTINCTION (A LOT)
- reflects absolute amount
- can add, subtract, parametric test
- can tell how much more or less a subject has
Examples:
Corey found 14 Easter eggs including the 3 eggs that every child started with.
Interval
Twelve students in the class were Psychology majors; 9 were History majors
Nominal
The women ranked themselves from 1-10, representing least to most attractive
Ordinal
Chrissi went out on 4 dates last month
Ratio
T OR F: Reporting that a statement was made can be data even if the contents of the statement are not data
T
The scales of measurement differ in:
the amount of info conveyed, math used, stat test
When Dr. Cotting has people keep their hands in ice water while completing a challenging word puzzle in his laboratory, the people from whom Dr. Cotting collects data are:
participants
Dr. Davis is testing the effect of testosterone on memory. She injects some rats with 25 micrograms of testosterone and some rate with 50 micrograms of testosterone. This type of level is
quantitative
Another name for control variables is?
extraneous variables
T or F: Operational definitions are sometimes used to assign subjects to groups
T
Subjects
do not know that they are being measured/observed, not given permission
Participants
aware, have given permission that they are being measured/observed
DVS:
data and quatifiable
Types of levels of IV?
quantifiable
qualitative
presence v. absence
If we are exploring the relationship between gender and academic performance, we would use an?
Quasi-experiment
In which of the following techniques may you attribute the differences in one variable to be caused by differences in another variable?
experiment
T OR F: Even if the correlation between number of hours studied and grade on a exam was very high, you could not conclude that the number of hours studied affected the grade on the exam
T
Ranges for defining correlation
- 00-0.29 (low correlations)
- 30-0.69 (medium correlations)
- 70-1.00 (high correlations)
A scatterplot that goes from the lower left to the upper right of the graph represents a correlation coefficient of
positive
Ideally, you want at least —pairs of measurements for a correlation to be meaningful.
10
Causes of low correlations
truncated range outliers errors in original measurement sensitive to linear relationships lack of relationship between two variables
T or F: A correlation is a good statistic to tell you whether two measurements are related to each other but should not be used to predict one variable from another
F
T OR F: Outliers and truncated ranges produce abnormally high correlations
F
A correlation coefficient provides two pieces of info about the relationship:
strength and direction
Correlations are best with which type of data?
Interval or ratio (DV)
A correlation by itself, NEVER shows that one variable
CIA (causes, interprets, affects)
With quasi-experiments, you identify only with … scale because participants place themselves in their own categories
NOMINAL
The—-is a mathematical proof that shows that if you took a large number of random samples from a population and plotted the means of those samples, the distribution of means will form a normal (bell) curve
Central Limit Theorem
If a student wants to conduct a research project at MBC, she must obtain permission from….before the research is conducted.
IRB (Institutional Review Board)
The experimenter is responsible for the …of the participants both before and after the research
welfare
T OR F: When a participant consents to participate in a study, she is agreeing to remain in the study until it concludes
F
T or F: The size of the sample is more important in evaluating how representative a sample is (of the population) than the size of the sampling errors and bias
F
Research w/ animals
- must maintain healthy, comfortable living conditions that are subject to inspection by an external committee that includes at least 1 vet
- must minimize discomfort
- must dispose of animals humanely
IRB protects?
Participants first, then experimenter.
T or F: The animals and humans are very sensitive about being watched can result in changes in behaviors called reactive behaviors
T
Limitations of obseravational research is:
- the observer must wait for the behavior to occur
- the observer cannot replicate the observation at will
- the observer cannot control or manipulate conditions
To say that the dog was delighted when her owner came home is an example of which of the following limitations of observation
anthropomorphism
T or F: One advantage of surveys is that it is very easy to generate appropriate questions
F
A realistically good response rate to surveys is approximately – or more?
60%
Disadvantage of survey
- hard to generate
- cannot determine cause and effect
From this point on, a full interpretation of a correlation will include — statements.
4
T OR F: In Psychology, if a p value is greater that .05, we can conclude that the relationship is significant
F
T OR F: If SPSS reports that the p value is .000, we should report p<.000
F
It was reported that “…r(50)=.23, p=.02” What does 50 refer to?
Degrees of freedom
T OR F: You should always report the exact p-value if you have it
T, always room for error
T OR F: If chi square determines that a choice is independent on the objects being chosen, we would expect that the difference between the numbers of people choosing each object is large
F, evenly divided
T OR F” The type of data typically found in chi square tests is ratio
F, ordinal and nominal
One-sample chi square tests require at least…subjects in each expected frequency category.
5
T OR F: Non-parametric tests are not as likely to find an existing difference between groups as parametric tests because data in non-parametric tests are less precise than data in parametric tests
T
best method of control
within-subjects, everyone gets the same treatment (matchined sampling)
next best: random sampling
Method of control
- make things exactly the same
- make the procedures as equal as possible
- balance that which you cannot control to the degree possible
- statistical control
The t-test can be used when there are —levels of the IV
2
A two-groups t-test is a …..design, a paired samples t-test is a …..design.
Between, within
T OR F: A t-test can tell you both whether the difference is means is large enough to believe that it is real and the size of this difference
F, EFFECT SIZE
T OR F: if the r^2 between two variables is equal to .45, 45% of the variance in one variable is shared by the other variable
T
Ranges for effect size
- 00-0.49 = small
- 50-0.70=medium
- 70-infinity=large
df for t-test
BS= N-2
WS: N-1
T OR F:We name designs based on their dependent variables
F
In the terminology of this chapter a factor is the same thing as a ….
IV
In a WS design
each subject is exposed to every level of each IV
When drawing a design, each column represents
level of an independent variable
T OR F: When drawing a BS Design, each cell will contain the same four names
F
T or F: We always know that error is involved in an experiment, and we can calculate precisely how much error is involved
F
T or F: if you decrease the probability that a Type I error might take place in your experiment, you increase the probability that a Type II error will occure
T
The likelihood of a Type I error occurring in an experiment is called
Alpha
T or F: The probability of beta can be precisely determined
F
The probability of failure to find a difference when there actually is a real difference is called:
Type II Error
You should use a WS ANOVA when each participant gets ….level of the IV and there are ….levels of the IV
every, more than two
After conducting a WS ANOVA, how many different degrees of freedom value should you report in the first results statement?
2
If you conduct a WS Anova and find that the computed F does exceeds the critical value at the .05 level, how many results statements should you provide?
3
In conducting a Tukey’s post hoc test, how many p values must you interpret if there are four levels in your study?
6
In which design are participants least likely to respond to demand characteristics of an experiment?
BS Design
For which design does a loss of a participant result in a greater loss of data?
WS Design
T OR F: There are 3 null hypothesis in a two-factor design
T
How many F ratios will be calculated in a two-factor design?
3
T OR F: If an interaction between 2 IVs is found to be significant, you still must analyze the main effects of each IV
F
T or F: Differences associated with levels of an IV inabsence of an interaction are called simple effectd
F
T or F: A significant interaction is indicated when the lines indicating the levels of one IV are parallel on a graph
F
Which two stats are most often depicted together on a graph of experimental results?
mean, standard error of the mean
The …variable goes on the y-axis
ordinate) and the …..variable goes on the x-axis (abscissa
Dependent, independent
Each line on a graph of a multi-factor design represents an
level
If there is no differece, but you find one
Type II error
If there is a difference, but you can’t find one
Type I error
