Stats

Question 1

What kind of data does a Chi squared test work with?

Answer 1

Usually nominal, but can do ordinal.

Question 2

Four different kinds of data

Answer 2

1. Nominal - labels, no ranks
2. Ordinal - categories with ranking
3. Interval - true, quantitative measures
4. Ratio - physical properties with no true value

Question 3

Three observations for a Chi squared test

Answer 3

1. Mutually exclusive classification - no one participant can belong to more than one category
2. Exhaustive categories - all members are accounted for
3. Independence of observation - each count is independent of one another

Question 4

What does contingency Chi tests look for?

Answer 4

Do the observed frequencies reflect the independence of two qualitative variables?
Independence = knowledge of the value of one variable tells us nothing about the knowledge of another

Question 5

Why do we do post hoc analysis?

Answer 5

Having a significant Chi value tells us that there is an association, but we don’t know where that association stems from - we have to find the cells that are significant

Question 6

What is the residual?

Answer 6

The deviation of the observed from the expected frequency

Question 7

Why do we do standardised residuals?

Answer 7

The size of the deviation is related to the size of the sample, standardising it accounts for different sizes of different cells and gives us a relative contribution

Question 8

Why do we use adjusted residuals?

Answer 8

Because standard residuals have a standard deviation of less than one, so we have to adjust them to get a proper idea of variance

Question 9

How do we interpret adjusted residuals?

Answer 9

They’re like z scores. For a = 0.05, anything bigger than 1.96 or smaller than -1.96 is significant. A positive residual says that there were more observed than expected, a negative residual says that there were less observed than expected

Question 10

How do we get from an adjusted residual to a p value?

Answer 10

1- NORM.S.DIST(abs(adjres))

Question 11

What kind of data is used in a t test?

Answer 11

Continuous data, usually interval

Question 12

Three assumptions of t tests

Answer 12

1. Normality - the sample tested come from a population that is normally distributed
2. Homogeneity of variance - pooled vs not pooled variance (k=4)
3. Independence - samples being compared do not influence one another (except for paired t-test)

Question 13

Decision making errors

Answer 13

Type 1 error - a false positive, falsely rejected the null hypothesis
Type 2 error - a false negative, falsely rejected the alternate hypothesis

Question 14

What is alpha?

Answer 14

The probability of making a type 1 error

Question 15

When do we use a one sample t test?

Answer 15

When comparing a sample to a known population, when population variance is unknown

Question 16

When do we use an independent sample t test?

Answer 16

• Comparing two samples
• Samples are separate and independent, with different subjects
• Sample sizes can be different

Question 17

Heterogenous vs homogeneous cut off

Answer 17

When the larger variance/smaller variance is less than 4, it’s homogenous. 4 AND greater, is heterogenous

Question 18

When do we use repeated measures/paired sampling t tests?

Answer 18

• Matched sampling units or repeated measures on the same group of participants
• when your assumption of independence is violated