Research and Program Evaluation Flashcards
True variance/coefficient of determination
Obtained by squaring the correlation
(number between 0 and 1 that measures how well a statistical model predicts an outcome.)
Construct Validity
When a test successfully measures a hypothetical construct.
What is most likely to threaten internal and external validity?
Selection of subjects
Ordinal scale data
Involves rankings or order of people/objects based on a particular attribute
ex: a horse categorized as a second-place winner in a race
Interval scale data
Calculated with the assumption that each number represents a point that is an equal distance from the point adjacent to it but NO zero point
can add and subtract but cannot multiply or divide
Ratio scale data
Have an absolute zero. (ex. weight)
Scatter plot
Depicts pairs of scores
Easy to see patterns (but broad generalization)
Instantly see positive or negative relationship between the data and if relationship is strong or weak
Purposeful sampling
NOT used to generalize the findings to the population (not random)
t-score
Mean of 50 and standard deviation of 10
Convergent validity
Convergent validity occurs when the construct being studied correlates highly with other constructs.
Content validity
The extent to which a measure represents all facts of a given social concept.
(ex. the condition of a sample from a larger population being a true representation of that larger population)
Ex Post Facto Study
Investigates possible relationships among variables after the fact.
Focus is on what has already occurred.
Also called casual comparative research.
Single blind vs double blind experiment
Single blind = subjects do not know whether they are in the placebo or experimental group
Double blind= neither the participants nor the experimenter know which subjects are in the experimental and placebo groups
internal validity
vs
external validity
internal= whether dependent variables were truly influenced by the experimental independent variables OR if other factors had an impact
external= whether the experimental research results can be generalized to larger populations
parsimony
interpreting the results in the simplest way
a test of significance
needed to compare a control group to the experimental group
p= probability
accepted level = .05 or less
Type I (ALPHA) error
vs
Type II (BETA) error
alpha error (I) = researcher rejects the null hypothesis when it is true
beta error (II) = researcher accepts null hypothesis when it is false
increased sample size
will lower the risk of chance/error factors
t-test
used to test for significant differences between groups
ANOVA
one-way analysis of variance is used when there is more than one level of single IV
ex: 1rst group receives no assertiveness training
2nd group receives 4 sessions of training
3rd group receives 6 sessions of training
with two IV would need a two way ANOVA or MANOVA
correlation coefficient
when a researcher uses correlation then there is no direct manipulation of the IV- nothing is manipulated just measured
a statistic that indicates the degree or magnitude of relationship between two variables
often abbreviated to lower-case r
a positive correlation is NOT a stronger relationship than a negative one (a - only indicates the fact that as one variable goes up the other goes down)
correlation does not mean casual
mode
most frequently occurring score and the least important measure of central tendency
in a basic curve the point of maximum concentration
regardless of the shape, will always be the high point when a distribution is displayed graphically BECAUSE it is the point where the most frequently occurring score falls
factorial experiment
used if a researcher wants to ferret out the effects of more than one IV
several experimental variables are investigated and interactions can be noted
bar graph is also called a….
histogram
x axis
used to plot the IV scores
also called the abscissa
a horizontal line is drawn nuder a frequency distribution
horizontal axis
y axis
also called the ordinate
used to plot the frequency of the DV
nominal scale
qualitative
simplest type of scale
used to distinguish logically separated groups
a DSM diagnostic category
no true zero point and does not indicate order
Hawthorne effect
if subjects know they are part of an experiment or if they are given more attention because of the experiment their performance sometimes improves
Rosenthal effect
experimenter expectancy effect
analysis of covariance technique
controls for sample differences which exist
helps remove confounding extraneous variables
statistically eliminates differences in average values influenced by covariates
demand characteristics
relates to any bit of knowledge (correct or incorrect) that the subject in an experiment is aware of that can influence their behavior
ERIC
educational resources information center
