Continuous Variables

Question 1

Descriptive statistics goals?

Answer 1

Check errors / outliers
Describe & summarize
Spread of data
Appropriate analysis decision
Parametric / Non-parametric

Question 2

Summarizing ratio or interval data? (3)

Answer 2

Central tendency
Dispersion
Normal curve, Skewness, Kurtosis

Question 3

Steps for inferential statistics?

Answer 3

Set null / alternative hypothesis
Est level of statistical significance (alpha)
Determine statistical significance (p value) of findings
Accept/reject null hypothesis

Question 4

Difference between Parametric & Non-Parametric tests

Answer 4

Para: assume data drawn from normally distributed population (data not skewed) and have same variance on variables being measured
Non-para: No assumptions about sample characteristics regarding distribution

Question 5

Describe the four types of variables

Answer 5

Interval: equal interval scale NO true zero
Ratio: interval scale with TRUE ZERO
Nominal: Male/female
Ordinal: Ranked data

Question 6

Describe a paired design

Answer 6

Each individual is own control (variation is limited)

Question 7

Determine non-parametric tests to use for Paired or Independent samples design

Answer 7

Paired: Wilcoxon signed rank test
Independent: Mann Whitney U test

Question 8

Difference between Independent & Paired data?

Answer 8

Independent: data from different groups of people, participant in ONE GROUP, differences between groups
Paired: data from group of individuals, collected from individual at different time points/conditions, differences between time points/cond

Question 9

Assumptions for Independent Samples t-test? (5)

Answer 9

DV: Ratio/interval
IV: two categories
Cond 1 & 2 measurements are independent
Variance of DV should not be very different
If n = <30, distribution of DV for each group should not be badly skewed

Question 10

Assumptions for Paired Samples t-test? (4)

Answer 10

DV: ratio/interval
IV: two categories
Each measurement in cond 1 has match in cond 2
Diff score obtained by deducting measurements

Question 11

What’s difference between Non-parametric to t-tests? (2)

Answer 11

Compares MEDIANS

Less powerful

Question 12

Assumptions for non-parametric tests? (4)

Answer 12

Assumptions of parametric tests breached (e.g. level of measurement, normal distribution, homogeneity of variances across groups)
Unable to correct for problems with data set distribution

Question 13

Assumptions for Mann-Whitney U test

Answer 13

Data must meet requirement that two samples are independent
Compare ranked values
Ranks assigned relative to both samples combined

Question 14

When to use Mann Whitney U test?

Answer 14

Sample sizes small, normality questionable
Data contains outliers, distort mean values affecting comparison
Data is Ordinal
Distributions of two groups are same shape
Not too many ties in ranks of data

Question 15

Differences between Sign Test and Wilcoxon Sign Ranked test?

Answer 15

Sign: differences between each variable, non para alt to one sample t-test
Wilcoxon: compare paired data, non para alt to paired t-test
Both used when normality cannot be assumed
Sign test count number of differences that are positive and negative