Content Review and Comparing Three or More Groups

Question 1

Are t-tests only in inferential statistics?

Answer 1

Yes
Same with comparing two 95% confidence interval and 95% confidence interval around a mean difference
*all can be used for comparing two groups and evaluating if they are significantly different

Question 2

What are f-tests?

Answer 2

They are used when comparing three or more groups in inferential statistics

Question 3

What is the goal of descriptive statistics?

Answer 3

To describe the data
Uses mean, median, and mode (measures of central tendency)
Uses standard deviation and interquartile range (measures of variability)

Question 4

What are inferential statistics?

Answer 4

Trying to infer information about the population based on the sample that we’re studying

Question 5

Why are descriptive statistics important?

Answer 5

Usually the first step to data analysis…inferential statistics rely on the “typical value” and “variability” for calculations
Essential to the critical review process because it gives you (as the reader) an overview of the dataset

Question 6

When do we use the mean?

Answer 6

When the data is not skewed
Normal distribution

Question 7

When should you use the median?

Answer 7

When the data is skewed and/or there are outliers

Question 8

When should you use the mode?

Answer 8

When using categorical or nominal data

Question 9

What is the standard deviation usually reported with?

Answer 9

The mean

Question 10

What is the interquartile range usually reported with?

Answer 10

The median

Question 11

Why are inferential statistics important?

Answer 11

Allows us to identify statistically significant differences
E.g. which intervention is most effective

Question 12

How do you infer?

Answer 12

To derive by reasoning
To reach a conclusion based on the evidence
To guess that something is true based on the information you have

Question 13

What is the basis of inferential statistics?

Answer 13

Infer something about the population based on what you are collecting within the sample(s)
Done by theoretical concepts and underpinnings of inferential statistics
Normal distribution, empirical rule, central limit theorem

Question 14

What is the empirical rule? How is it related to normal distribution?

Answer 14

68% of the observations fall within 1 standard deviation of the mean
95% of the observations fall within 2 standard deviations of the mean
99.7% of the observations fall within 3 standard deviations of the mean
In a normal distribution, most values lie within 1 SD of the mean and almost all values lie within 2 SD of the mean

Question 15

What is a sampling distribution?

Answer 15

Distribution of a statistic over a set of theoretical samples
Distribution of sample means
The mean of the sampling distribution is the mean of the population
Would get a normal distribution if you plot enough sample means

Question 16

How do you know if significant group differences exist?

Answer 16

By performing tests
t-test (independent or dependent)
Distribution of many sample means
Theoretical distribution
Two 95% confidence intervals
Comparing 95% confidence intervals around the mean difference

Question 17

Does the sampling mean allow us to infer things about the population?

Answer 17

Yes

Question 18

When is there a significant difference between values for a t-test? (both independent and dependent)

Answer 18

When the t-score is greater than the t-value

Question 19

When are there significant differences when comparing two 95% CI? (both independent and dependent)

Answer 19

When the two 95% CIs do not overlap (or do NOT share a value) with each other

Question 20

When are there significant differences when comparing 95% CI around the mean difference? (both independent and dependent)

Answer 20

When the 95% CI around the mean difference does NOT include “0”

Question 21

How do you conduct a t-test?

Answer 21

Find the t-value (critical value), which BASED ON dof & t-table
Refer to the t-table
Use the t-table to find the critical value
Degrees of freedom is (n-1)
n = sample size (number of participants)
Calculate the t-score, which is BASED ON YOUR SAMPLE
To calculate the t-score, you need the mean and standard error of the mean
Compare the t-score to the t-value (critical)
If the t-score is greater than the t-value, you have significant differences
If the t-score is less than the t-value, you do NOT have significant differences

Question 22

How do you calculate your t-score for independent t-test?

Answer 22

Mean 1-mean 2/standard error of the mean difference

Question 23

How do you calculate the 95% CI around the mean difference?

Answer 23

Mean 1 - mean 2 +/- t-value (standard error of the mean difference)

Question 24

How do you compare two 95% CIs?

Answer 24

mean +/- t-score (standard error of mean)

Question 25

Does comparing 95% CIs require the standard error of the mean difference?

Answer 25

No So if you are not given that value, you should be comparing the 95% CIs CI for each group calculated separately and examining if they overlap

Question 26

How do you calculate the standard error of the mean?

Answer 26

Standard deviation of the sample/square root of sample size

Question 27

How do you calculate the t-score for dependent t-tests?

Answer 27

mean of the difference in scores/standard error of difference in scores

Question 28

What do you do if you are not given the standard error of the difference in scores?

Answer 28

Calculate it Standard deviation/square root of the sample size

Question 29

How do you calculate the 95% CI around the difference for a dependent t-test?

Answer 29

Mean of the difference in scores +/- t-value (standard error of the difference in scores)

Question 30

What test do you need to do if there are three of more test conditions?

Answer 30

RMANOVA (one group) ANOVA (multiple groups)

Question 31

Are ANOVA and RMANOVA the same thing as an f-test?

Answer 31

Yes

Question 32

What is the process of an f-test

Answer 32

Find the F-value (critical value), which is BASED on dof and F-table Calculate the F-score which is BASED ON YOUR SAMPLE For our class, the F-score will be given to you **No calculations for this class** Compare the F-score to the F-value (critical)

Question 33

How do you find the degrees of freedom for an f-value?

Answer 33

Degrees of freedom #1 is the number of groups or conditions in the study (k-1) (ie, the columns in table); k = number of groups/conditions Degrees of freedom #2 is the number of subjects across groups (n-k) (ie, the rows in table); n = sample size (total number of participants) Find the value on the f-table

Question 34

How do you compare the f-score to the f-value?

Answer 34

If the F-score is greater than the F-value, you have significant differences If the F-score is less than the F-value, you do NOT have significant differences

Question 35

Are all ANOVA tests omnibus tests?

Answer 35

Yes All have one major limitation - it doesn't tell you where the differences are between different groups Does not tell you which intervention is the best Need to take a step further and run multiple comparisons test

Question 36

How should you follow up an omnibus test?

Answer 36

Multiple comparisons using multiple t-tests Resetting the alpha level will protect against making a Type I Error - needs to be reset between each t-test

Question 37

What is the typical alpha level?

Answer 37

0.05 This will have to be lowered to protect against for type I errors The Bonferroni correction is the most basic approach to reset the alpha level

Question 38

What is the bonferroni correction?

Answer 38

alpha level/number of comparisons *needs to be done when you are making multiple comparisons; not just 1v1

Question 39

What does the alpha level mean?

Answer 39

Anything below the alpha is significant