Stats Flashcards
Alpha
probability of rejecting the null when it is true
ANCOVA
statistically removes variablility in DV due to EV
Normal distribution
1SD = 68, 2SD = 95, 3SD = 99
Assumptions of Pearson r
linear, unrestricted range of scores, homosc (underest)
homoscedasticity
equal variablility of Y scores at all values of X
Canonical Correlation
cor of 2 linear combos of variables, predicts stat on criteria
Central limit theorem
samp dist of mean approach normal as sample size grows
cluster sampling
selecting groups from population: schools, hospitals
coefficient of determination
r2. Amount of var in Y accounted for by variability in X
correlation coefficient
relationship between two variables
counterbalanced design
administering different levels of IV in different orders
cross-sequential
combines sross-sectional and longitudnal
discriminant function analysis
2 or more continuous predictors, one discrete criterion
event sampling
rare behaviors or leave a permanent record
Internal validity
variability is due to the IV
interval recording
divide time into intervals, record if beh. Occurred
kurtosis
peakedness of distribution.
leptokurtic
tall distribution
platokurtic
flat distribution
Least squares criterion
stat used to locate reg. Line to reduce error
lisrel
verifies a predfined model or theory
Mann-Whitney U
non parametric, 2 independent groups
Mean squares between
numerator of F ratio, variability due to IV and error, MSB is larger,
Mean squares within
the denominator of the F ratio, variability due to error
Mulicollinearity
bad state for MV: predictors are highly correlated
Multiple baseline design
appying treatment to different behaviors, settings, groups.
Multiple correlation coefficient
R. Relat between 3 or more variables
Multiple regression
predicts score on continuous criterion
ANOVA
one IV and one DV, each interval or ratio. Preferable to multiple t tests
Path analysis
verify a predefined causal model or theory.
Scheffe
post hoc, least risk of type I, pairwise and complex
Tuckey
post hoc, least risk of type I, pairwise when equal size groups
Fisher LSD
post hoc, least risk of type II, pairwise and complex
Power
probability of rejecting a false null
Protocal analysis
Have subject think outloud and analyze the record of verbalizations
randomized block fact. anova
EV is treated as another IV, can analyze main and interaction effects of EV
reactivity
effect of being aware of being in a study
regression analysis
Predicts a score on a criterion based on a predictor
rejection region
sample means that are unlikely to be the result of sampling error, = alpha
reversal withdrawal design
ABA
Solomon Four
Pretest is considered an IV
effects of adding to raw score
+or - changes CT not SD, * or / changes CT and SD
Statistical regression
scores moving toward mean when retested
Latin square
each level of the IV appears the same number of times in each position
Factorial design
two or more Ivs
Sampling error
how sample differs from population
maximize power
Increasing alpha, big sample, IV more intense
Confidence
increases as alpha decreases
single sample chi square
one variable, many categories
multiple sample chi square
2 ore more variables, many categories
assumption of chi square
observations are independent
Wilcoxen matched pairs
2 correlated groups, alternative to t-test for correlated samples
Man whitney U
2 independent groups, alternative to t-test for independent samples
Kruskal wallace
2 + independent groups, alternative to one-way ANOVA
Parametric
assumption: normal distribution, homoschedasticity
t-test for single sample
compare sample mean to known mean
t-test for correlated samples
two correlated groups (a single group is compared to itself)
t-test for independent samples
two independent groups (experimental and control)
One way Anova
1 IV , 1 DV, and 2+ independent groups
Factorial Anova
2 or more IV (2 way = 2 Ivs). If sig, analyze main and interaction
Manova
1 or more IV and 2 or more DV
Pearson r
both variables are interval or ratio
spearman rank order
both variables are rank ordered
Phi
both variables are true dichotomies
Tetrachoric
both are artificial dichotomies
contingency
both are nominal
Point biserial
one is true dichotomy and one is interval or ratio
biserial
one is artificial dichotomy and one is interval or ratio
Eta
assess non-linear relationships; both are interval or ratio