Use LEFT and RIGHT arrow keys to navigate between flashcards;
Use UP and DOWN arrow keys to flip the card;
H to show hint;
A reads text to speech;
67 Cards in this Set
- Front
- Back
E.g. nominal scale
|
sex
eye colour DSM diagnosis religion i.e. unordered categories |
|
E.g. ordinal scale
|
likert scale
i.e. categories and rank order |
|
E.g. interval scale
|
IQ scores
temperature i.e. rank order and equal intervals can add & subtract |
|
What type of measurement?
1. religion 2. IQ score 3. DSM diagnosis 4. height 5. gender 6. likert scale 7. reaction time 8. frequency of agressive acts |
1. nominal
2. interval 3. DSM diagnosis 4. interval 5. nominal 6. ordinal 7. ratio 8. ratio |
|
E.g. ratio scale
|
# calories consumed
reaction time i.e. rank order, equal intervals AND absolute zero can multiply & divide |
|
Types of measurement in increasing complexity
|
N ominal
O rdinal I nterval R atio |
|
Cumulative frequency
|
total number observations that fall at or below each score
|
|
Kurtosis
|
the relative peakedness of a distribution
more peaked = leptokurtic more flat = platykurtic |
|
Skewed distribution
|
more than half observations fall on one side of distribution
postive = low score negative = high score |
|
Measure of central tendancy
|
mode (Mo)
median (Md) mean (M or X) |
|
Relationship bet measures of central tendency in skewed distributions
|
positively skewed d:
mean, greater than median, greater then mode negatively skewed d: mode, greater than median, greater than mean |
|
Measures of variance
|
range
variance standard deviation |
|
Def: variance
|
a measure of variability calculated by dividing the sum of squares
SS / n (population) SS / n-1 (sample) |
|
Def: standard deviation
|
square root of the variance
a measure of variability of scores around the mean SS / n - then take square root |
|
Re: Inferential statistics,
What is a sample statistic used for |
to estimate a population parameter
|
|
Sampling error
|
random error responsible for diff bet sample values and population
|
|
Sampling distribution of the mean
|
distribution of means obtained if large number of equal-size random samples are drawn from same population
|
|
3 predictions of Central Limit Theorem
|
1. as sample size increases, sample distribution of mean approaches normal distribution
2. mean of sampling distribution of the mean = population mean 3. SD of sample distribution of the mean = population SD divided by square root of sample size |
|
Def: Standard error of the mean
What does it measure |
SD of sampling distribution of the mean
variability due to effects of random error |
|
What happens to standard error when
1. SD larger & sample size smaller 2. SD samller & sample size larger |
1. SE larger
2. SE smaller |
|
2-tailed vs 1-tailed hypothesis
|
2-tailed = nondirectional
1-tailed = directional |
|
rejection / critical region
|
region of unlikely values
lies in one or both tails of sampling distribution values occur as a result of sampling error |
|
retention region
|
region of likely values
lies in central portion of sampling distribution |
|
What happens to hypotheses if sample statistic is in rejection region
|
null hypothesis is rejected
alternate hypothesis is retained |
|
What happens to hypotheses if sample statistic is in retention region
|
null hypothesis is retained
alternate hypothesis is rejected |
|
Size of rejection region defined by...
|
alpha
level of significance note: alpha = 0.05 means 5% in rejection region |
|
Type I error
|
reject a true null hypothesis
|
|
Type II error
|
retain a false null hypothesis
|
|
Statistical "confidence"
|
certainty about the decision re: null hypothesis
|
|
Statistical "power"
|
ability to reject a false null hypothesis
|
|
Parametric tests
|
evaluate hypotheses about population means, variances etc.
e.g. t-test, ANOVA interval or ratio scale |
|
non-parametric tests
|
evaluate hypotheses about shape of distribution
e.g. Mann-Witney U test, Wilcoxon ordinal or nominal scale |
|
degrees of freedom
|
N-1 (t-test)
C-1 (chi-square) |
|
What information do you use to select an inferential statistic?
|
scale of measurement
dependent variable study design |
|
What statistical test would you use for nominal data?
|
single-sample Chi-square
multiple-sample Chi-square |
|
What statistical test would you use for ordinal data?
|
Mann-Witney U-test
Wilcoxon matched pairs test Kruskal-Wallis test |
|
What are the non-parametric alternatives to:
1. independent t-test 2. correlated t-test 3. one-way ANOVA |
1. Mann-Witney
2. Wilcoxon 3. Kruskal-Wallis |
|
What statistical tests would you use for interval and ration data?
|
t-test
ANOVA |
|
Name types of t-test
|
simple sample
independent samples (between) correlated samples (within) |
|
Name types of ANOVA
|
one-way
factorial (2-way, 3-way) randomized block factorial ANCOVA repeated measures mixed (split-plot) MANOVA |
|
When use one-way ANOVA vs. factorial ANOVA?
|
one-way = 1 IV
factorial = 2+ IVs |
|
What are the Post Hoc tests for ANOVA?
|
Scheffe's S test
Tukey's HSD test Fisher's LSD test |
|
Which Post Hoc test is least vulnerable to Type I Error, but more vulnerable to Type II error?
|
Scheffe's
Tukey's |
|
Which Post Hoc test is least vulnerable to Type II Error, but more vulnerable to Type I error?
|
Fisher's
|
|
The numerator of the f-ratio is a measure of variablity due to...?
|
treatment effects & error
|
|
In ANOVA, the "mean square within" provides info about:
|
sampling fluctuations
|
|
Why use one-way ANOVA instead of seperate t-tests?
|
to reduce Type I error rate
|
|
How do you calculate f-ratio?
|
MSB/MSW
|
|
How do you calculate MSB?
|
SSB/df
|
|
How do you calculate MSW?
|
SSW/df
|
|
Use: MANOVA
|
1+ IV
2+ DV (interval/ratio) *helps increase statistical power by assessing effects of IV on all DVs |
|
Ex: planned "a priori" analysis (4)
|
Dunn-Bonferroni t
linear contrasts orthogonal comparisons trend analysis |
|
Axis on scattergram
|
X = IV = predictor
Y = DV = criterion |
|
Which correlation coefficient is most commonly used with...
1. interval and ratio data 2. rank data 3. nominal data |
1. Pearson r (also Eta)
2. Spearman rank 3. Contingency (C) |
|
How do you translate correlation coefficient score into something meaninful?
|
calculate coefficient of determination to provide a measure of shared (explained) variability.
- square the coefficient e.g. if coefficient is .60, .60 x .60 = .36, .36 x 100 = 36%, therefore, 36% of scores on DV explained by IV... remaining 64% is unexplained variance. |
|
Use: regression analysis
|
to predict a score on a criterion (DV) based on person's obtained score on predictor (IV).
|
|
How do you locate a regression line
|
least squares criterion
|
|
Use: multiple regression
|
2+ continuous or discrete predictors
1 criterion |
|
Ex: multiple regression
|
1. simultaneous (simple)
2. stepwise 3. hierarchical |
|
When use multiple regression instead of ANOVA?
|
if groups are unequal in size
if IVs on a continuous scale |
|
Use: canonical correlation
|
[an extension of mult reg]
2+ continuous predictors 2+ continuous criterions |
|
Ex: multivariate techniques (4)
|
multiple regression
canonical correlation discriminant function analysis logistic regression |
|
Ex: bivariate correlational techniques (2)
|
scattergram
correlation coefficient |
|
Ex: bivariate prediction (1)
|
regression analysis
|
|
Ex: multivariate techniques (2)
|
path analysis
LISREL |
|
Ex: correlation & prediction tehcniques (4 main)
|
bivariate correlational techniques
bivariate prediction multivariate techniques: correlation & prediction multivariate techniques: causal modeling |
|
Use: discriminant function analysis
|
2+ continuous predictors
1 discrete criterion |