Stats/Research Design Flashcards
MANOVAs are for how many DVs
more than one
One way ANOVAs are for how many IVs
one
Factorial ANOVAs are for how many IVs
more than one
Which is more powerful - ANOVA or MANOVA?
ANOVA
It’s easier to find significant if separate ANOVAS are run for each IV but this increases the chance of _____
Type 1 error
ANCOVA is used for removing the effects of a
Confound
Selection bias is a threat to _____ and occurs when….
Internal validity….subjects are not randomly assigned to tx groups
Quasi-experimental designs affect internal validity, external validity, both, or neither?
internal validity only
Is external validity affected by non-random assignment?
No
Experimental is classified as either ____ or ____
True or Quasi
What is the primary feature that distinguishes a true experimental study from a quasi one? Is it considered the most powerful method of experimental control?
Random assignment…YES
A source of systematic error that is irrelevant to a study’s hypotheses and confounds the results
extraneous variable
Some technique to control for an extraneous variable include:
1) Random assignment to tx groups
2) Holding the EV constant
3) Matching subjects on the EV
4) Building the EV into the study (aka Blocking)
5) Statistical control of the EV
What is the primary downfall of holding an EV constant to control for its effects?
Limited generalizability
Matching subjects on the EV is most useful when?
The sample size is too small
When Building the EV into the study (aka Blocking), the EV is then considered a?
Moderator variable
In choosing a research design it is critical you choose one that minimizes the effects of ______ and _____ error.
systematic, random
Internal validity is threatened when which 3 things happen?
1) Investigator can’t maximize the effects of the IV
2) Investigator can’t control the effects of any EVs
3) Investigator can’t minimize the effects of random error
Name threats to internal validity
1) Maturation - changes within subjects over time
2) History - comes from out there
3) Testing
4) Instrumentation - tests themselves change in accuracy or sensitivity
5) Statistical Regression
6) Selection
7) Attrition
8) Interactions with Selection
___ validity is always limited by its ____ validity
external limited by internal
if you content there’s a causal relationship between variables in the context of the study, you _____ conclude that there’s a relationship for other people or circumstances
CANNOT
A high degree of internal validity _____ guarantee external validity
DOES NOT
the sd is the square root of the _____
variance
variance squared is the _____
sd
A threat to internal and external validity is _____, which is when participants react to an IV in a certain way because they know they’re being observed
reactivity
Three types of reactivity:
1) Evaluation Apprehension
2) Demand Characteristics
3) Experimenter Expectancy
When participants behave in a certain way to avoid negative evaluations
Evaluation apprehension
When the experimenter unintentionally provides subjects with cues about what’s expected
Experimenter expectancy
Cues in the experimental setting that inform subjects of the purpse of the study
Demand characteristics
A threat to internal and external validity occurs when is pretest sensitization which can be controlled by ether of the following two
1) Solomon 4 group design (pretest is considered an IV)
2) Don’t administer a pretest
____ interference can be controlled by using a counterbalanced design, which means all subjects receive the treatment in ____ order
Multiple treatment, different
Multiple treatment interference can have _____ or ____ effects
order, carryover
A type of counterbalanced design
Latin square
Carryover and Order effects are common in _____ interference
Multiple treatment
When marginal means are different, one can _____ conclude that there are ____ effects
tentatively, main
A study can have _____ effects even if there is no ___ effect
interaction, main
There can be an combination of interaction and main effects. T or F
T
A set of data that uses ordinal, interval, or ratio scales can be organized using a frequency _______
polygon
Measures of central tendency
Mean, Median, Mode
If one score in a distribution is extremely high or low, the value of the ____ in not affected
median
of the 3 measures of central tendency, the ___ is least susceptible to sampling fluctuations
mean
the most susceptible to sampling fluctuations of the 3 measures of central tendency is the ____
mode
What are the best measures of central tendency for the following scales of measurement:
1) Nominal
2) Ordinal
3) Interval
4) Ratio
1) Nominal - Mode
2) Ordinal - Mode or Median
3) Interval - Mode, Median, Mean
4) Ratio - Mode, Median, Mean
These measures indicate the amount of heterogeneity or dispersion within a set of scores:
measures of variability
Name 3 measures of variability
1) Range
2) Variance
3) Standard Deviation
percentage of scores that fall between +/- 1 sd
68%
percentage of scores that fall between +/- 2 sd
95%
percentage of scores that fall between +/- 3 sd
99%
the # of cases that fall within limits that are defined by the sd - these limits are referred to as
areas under the normal curve
The central limit theorem states that the ____ mean is equal to the ____ mean and the sd of the sampling distribution is equal to the population sd divided by the ____ of the ____ size.
sample, population, square root of the sample size
The smaller the ____ size the ___ the standard error of the mean
smaller, larger
Formula for Variance
SS/N-1
To calculate the SS, you
subtract each score from the mean, then sq each, then add them all
S squared =
variance
Formula for SD
Square root of SS/N-1
The population mean is represented by a symbol that looks like __, while the sample mean looks like ___ or____
u
X or M
The population standard deviation is represented by a symbol that looks like __, while the sample standard deviation looks like ___ or ___
o
S or SD
The population variance is represented by a symbol that looks like __, while the sample variance looks like ___ or ___
o squared
S squared or V
Formula for the Standard Error of the Mean
o/sq rt of N
For there to be able to say there is an effect or the effects are probably due to IV variance and not due to error, you want your scores to fall within the ___ region and not the _____ region of a distribution
rejection, retention
If a score is in the rejection region, the null hypothesis is
rejected
The size of the rejection region is defined by
alpha
a Type __ error, which is ____ a true null, is defined by Alpha
I, rejecting
A Type __ error, which is ____ a false null, is defined by Beta
II, retaining
Power, which is correctly ____ a false null, is defined by
rejecting, 1 - Beta
Correctly retaining a ___ null is defined by ___
true, 1 - A
Nonparametric tests are for ____ and _____ data
nominal and ordinal
Parametric tests are for _____ and _____ data
interval and ratio
When selecting which inferential statistic, you want to first consider which 2 factors?
1) Scale of measurement of data to be analyzed
2) Design of the study
the nonparametric version of the one-way ANOVA is the
kruskal-wallis
the nonparametric version of the t-test for independent samples is
mann-whitney U
the nonparametric version of the t-test for correlated samples is the
wilcoxon matched-pairs test
2 assumptions of parametric tests are that the value of interest is _______ in the population. When the study includes more than 1 group there is _______.
normally distributed, homoscedasticity
the variances of the populations the different groups represent are equal is referred to as ______
homoscedasticity
A violation of the parametric test assumptions, especially _______, increases the chance of which error or errors?
homoscedasticity, Type I and II
Parametric tests are considered _____, which means that some deviation from a normal curve or homoscedasticity won’t invalidate results
robust
Some ways to maximize robustness are to:
1) Equal number of subjects in each group
2) Set alpha at a lower level
3) Increase sample size
Because nonparametric tests don’t rely on a normal distribution, they’re referred to as _____ free tests
distribution free
______tests are used to evaluate the ___ of a distribution rather than its mean, variance, or other parameter
nonparametric, shape
F-ratio =
MSB/MSW
mean squares between/mean sqaures within
Effect size can be represented by
Cohen’s d
Eta squared
A high correlation between 2 or more predictors is referred to as _________, and IS or IS NOT desirable..why?
multicollinearity, IS NOT - means it is difficult to tell which predictor leads to change
