N211 Stats Review Flashcards
Independent samples are ____ while dependent samples are ____
unrelated, related
Between parametric and non-parametric tests, which are more powerful?
Parametric
Between parametric and non-parametric tests, which are more likely to make a Type 2 Error?
Non-parametric
The non-parametric equivalent of an independent t-test is:
Mann-Whitney U test
The non-parametric equivalent of a paired t-test is:
Wilcoxon Signed-Rank test
The non-parametric equivalent of a one-way ANOVA is:
Kruskal-Wallis test
The non-parametric equivalent of a one-way repeated measures ANOVA is:
Friedman’s ANOVA
What are the 3 requirements for using a chi-square test (sample, data, value)?
- 2 random independent samples
- Data is nominal or ordinal
- Value of each cell must be >5
What is the null vs alternate hypothesis for chi-square?
Ho: no difference between 2 groups
Ha: significant difference between 2 groups
How do you calculate degrees of freedom for chi-square tests?
Df = (# of rows - 1) x (# of columns - 1)
If the X^2 sample is > the critical value, is the result statistically significant? Do you accept or reject the Ho?
Yes, reject Ho
What are the 4 requirements for using t-tests (sample, dv, variance)?
- 2 random samples
- Interval or ratio DV
- Normal distribution of DV
- Homogeneity of variance
How do you calculated degrees of freedom for t-tests?
Df = (sample size of both groups) - 2
For t-tests, is p > or < alpha if Levene’s F value is significant? Does this mean we assume equal variance? Do we accept or reject Ho?
p < alpha, DON’T assume equal variance, reject Ho
If t sample is < t critical value, is it significant? Do you accept or reject Ho?
Not significant, accept
ANOVAs compare how many sample means?
2+
What is the null and alternative hypothesis for ANOVAs?
Ho: sample mean 1 = 2 = 3
Ha: sample means aren’t equal
What are the 4 requirements for using ANOVAs (sample, LOM, variance, distribution)
- Random, independent samples
- Interval or ratio LOM
- Homogeneity of variance
- Normal distribution
How do you calculated degrees of freedom for ANOVAs?
Df = df numerator + df denominator
> Df numerator = # of groups - 1
> Df denominator = total # of subjects - # of groups
What is the formula to calculate F-ratio?
F-ratio = (difference b/w groups)/(difference within groups)
When an F-ratio is close to 1, do you accept or reject the Ho?
Accept: little difference b/w groups
If you don’t have a p value and only an F-ratio, what should you do?
Compare F sample against F critical value table
If F sample is > F critical value, is this significant? Do you accept or reject Ho?
Significant, reject
What are the 2 requirements for using a repeat measures ANOVA (sample, symmetry)?
- Dependent sample
- Compound Symmetry
What are 3 concerns with repeat measures ANOVAs?
- Position/latency effects
- Carryover effects
When is the spearman correlation coefficient used?
Ordinal OR not normally distributed
When is Pearson’s correlation coefficient used?
Interval/ratio AND normally distributed
Correlation coefficients describe what 2 things?
- Direction
- Strength of a linear relationship
When a correlation coefficient is positive vs negative, what does this mean in terms of direction?
Positive: variables move in the same direction
Negative: variables move in opposite directions
Describe the strength of the relationship:
p/r < 0.3 : ____
0.3 < p/r < 0.5 : ____
0.5 < p/r : ____
weak, moderate, strong
How do you calculate the percentage of variance?
r^2 x 100%
Percentage of variance describes the % of differences found in the ____ variable can be explained by the ____ variable
dependent, independent
Clinically, if r > __, it is significant
0.3
If 2 variables have a strong correlation, this means the effect size is ____ and the sample size is ____
large, small
What is the purpose of linear regressions?
Make predictions about the future value of DV at a given IV level
What is a residual?
Difference b/w actual data points & predicted data points
What is the slope of trendline?
How much DV changes per 1-unit change of IV
What is the formula for multiple regression?
y = a + b1x1 + b2x2 + e
What is the formula for odds ratio (OR)?
odds of outcome occurring/odds of outcome not occurring
What is the deference between sensitivity & specificity in terms of screening tests?
Sensitivity: increased true positives, decreased false negatives
Specificity: increased true negatives, decreased false positives
What is a positive predictive value (PPV)?
Probability of a true positive
What is a negative predictive value (NPV)?
Probability of a true negative
When prevalence increases, PPV will ____ meanwhile NPV will ____
increase, decrease
What is EFF (efficiency)?
How correct the screening test was vs actual diagnosis
