Flashcards
Statistics Theory Quiz
Cards in this deck (27)
- 1What is the difference between descriptive and inferential statistics?Descriptive: measuring trait of group without intention to generalise to public
Inferential: make generalisations from a smaller group to larger group (prediction) - 2What are the 3 types of descriptive statistics?-Central tendency (mean, median, mode)
-Spread of variation (s.d. variance)
-Distribution (shape of rates of occurence of values within group, normal, bimodal) - 3Explain 3 measures of central tendency:Median: middle of observations
Mode: most frequent value (only discrete data)
Mean: mathematical average - 4List the measures of spread (variability) in a distributionRange: min -> max values
Quartiles: 1st (25% data below this level)
3rd (25% data above this level)
Standard deviation: average deviation of scores from mean (small s.d. indicates data points close to mean, divide by n-1 instead of N in samples vs. pop.) - 5List two other measures of variabilityVariance: (s.d. ^2)
Coefficient of variation: compare variability of 2+ distributions whose measured variables are not the same (CV=100 s.d./mean) - 6What is kurtosis?-The description of the vertical characteristic of the curve (can still be normal distribution with high kurtosis)
-Positive kurtosis: sharp peak
-Negative kurtosis: blunt curve - 7What are the types of distributions?-Normal
-Positive & negative skewed distribution
-Bimodal distribution (two modes) - 8What are the 3 tests for normality & parametric testing?-Kolmogorov-Smirnov test
-Shapiro-Wilk
D'Agostino-Pearson - 9What is the Standard Error of the Mean (s.e.m.)?-The s.d. of the sample means
s.e.m=s.d./sq.r(n) - 10What is a confidence interval?-Expected upper & lower limit for true pop. mean estimated from sample
-CI=mean+-(s.e.m. x confidence level value)
-Means 95% of observations occur within these parameters which is 1.96 s.d. of mean - 11What does the p value mean?-P=1 means 100% certainty of even occurring
-P=0 mean 0% chance of event occurring
-p<0.05 means unlikely to be due to chance, due to correlation/causation - 12What is effect size?-Gauges meaningfulness of differences between two measures of central tendency
-ES=(mean G1 - mean G2)/s.d.
-s.d. pooled across two groups
-0.2=small ES, 0.5=moderate, >0.8=large - 13What is the Chi-Squared test for?-Analyse categorical data
-Tests sig. diff. between frequency of observed vs. expected observations - 14Explain odds ratio vs. relative riskOR: compares relative odds of event in group e.g. tell us less chance female dying
RR: compares incidence of an event in each group, rather than odds e.g.risk of female dying compared to male is 0.4, e.g. risk female dying 40% that of male - 15What is the null hypothesis?-no difference between two population means
Alternative hypothesis- logical state of reality if the null hypothesis is not true (i.e. if reject null hypothesis, accept alternative) - 16What causes statistical error?-TI - rejecting Ho when it is actually true, Probability of making error is alpha level
-TII - accepting Ho when false,
Probability of making error is beta level
-As alpha increases TI increased and TII decreased - 17What is statistical power?-Detecting real diff., probability of making correct decision
-P=1-beta
-Increasing sample size increases power
-^ES, decreases Beta, ^ Power - 18How do we calculate correlation?-Pearson's r
-R can be -ve of +ve
- <.25 is weak, .26-.50 is moderate, .51-.75 is fair, .76 is high
-As n increases, correlation required for statistical sig. decreases (i.e. as n increases, a lower r can be stat. sig.) - 19What is the r squared value?-Coefficient of determination
-Ratio - explained variation:total variation
-E.g. r^2 of 0.518 means 51.8% of variation in one variable is attributable to variation in the other - 20What is the co-efficient of non-determination?-K^2 = (1-r^2)
-Proportion of variance not explained in terms of correlation between X & Y - 21What are the types of tests used to test differences between groups (parametric)?-T-test
-ANOVA (analysis of variance)
-Effect of treatment or manipulated variable (IV) on an outcome (DV) - 22What are the underlying assumptions of parametric tests?1. Normally distributed
2. Random samples from population
3. Similar variance
4. Data is continuous - 23What are the different types of t-test?-One sample t-test: single sample & compare to value assumed to be pop. mean
-Independent t-test: two samples representing different groups, groups differ along one independent variable
-Paired t-test: Same subjects measured twice under different conditions - 24In which circumstances would you choose a one-tailed or two-tailed test?-Using a two-tailed test can always be justified
-One-tailed, only when know direction of results - 25When do we perform an anova?-want to compare multiple groups, 1 vs. 2 vs. 3hrs /wk
-power production in sprinters vs. jumpers vs. swimmers etc. (e.g. must have 4 different variables) - 26Why do we use ANOVA over multiple T-tests?-Probability of TI error is too great, ANOVA takes care of this.
- 27—
Information
- Description
- Mid-semester Statistics Theory HMST3846 - Research Skills Flashcards on Statistics Theory Quiz, created by Mikaela Sullivan on 30/04/2018.
- Created by
- Mikaela Sullivan
- Views
- 3
- Tags
- Rating