Lecture 15

Question 1

What is an interval scale? Example?

Answer 1

Interval scale: Ranked & intervals between values are meaningful

Example: Temperature in Celsiuso
30 is greater than 29
31 is as far away from 30 as it is from 29
10-point difference is meaningful
0 is NOT the absence of temperature***
We can compare arithmetic differences

Question 2

What is a ratio scale?

Answer 2

Ratio scale: Ranked, meaningful intervals, & absolute 0

Example: Age
- Ranked; 2 years is older than 1 year
- Intervals are equal
- 0 years is 0 years

Example: Reaction time in seconds
0 for reaction time, means simultaneous response
We can compare ratio differences

Question 3

What descriptive statistics do you use for nominal scale data? ordinal scale data? interval and ratio scale data?

Answer 3

Nominal scale data
- Only use mode for
central tendency

Ordinal scale data
- Can use both mode and
median

Interval & Ratio scale data
- Can use mode, median,
and mean

Question 4

What do the most common statistics require?

Answer 4

Most common statistics (e.g., t-test or ANOVA) require interval/ratio data

Question 5

What are likert type scales?

Answer 5

ordinal scales.

Question 6

Have you ever seen. mean calculated for Likert estimates?

Answer 6

Archival research of 2016 publications in top 4 journals (Liddell, 2017)

Percentage that calculated a mean?
100%!!

Question 8

How can ordinal scale data be treated with good justification?

Answer 8

With good justification, ordinal scale data may be treated as if it approximated interval-level data
It is wrong
Sometimes, the wrong-ness is worth it!

Question 8

What are the best practices for treating likert measures as if they were interval?

Answer 8

1. Look at the distribution of your data
2. Pre-test,or pilot-test your scale anchors
3. Analyze data both ways (i.e.,compare group means vs. group medians)
4. Tell your reader!

Question 9

When do you use bar graphs?

Answer 9

Bar graphs: Compare group means, group % § Bar graph of means

Independent variable: Nominal or ordinal

Dependent variable: Interval or ratio

Bar graph of percentages

Independent variable: Nominal or ordinal

Dependent variable: Nominal or ordinal

Question 10

When do you use box plots?

Answer 10

Box plots: Compare group medians, visualize density

Independent variable: Nominal or ordinal

Dependent variable: Interval or ratio

Violin plots, ”box’n’whisker”

Question 11

What is the effect size?

Answer 11

How strong is the relationship between variables?

Example: Relationship between emotion (sad vs. happy) and word
recall

Need a metric we can compare across different experiments and operationalizations of variables

Effect size = Difference between means / Variability

Magnitude of the relationship between group membership and the dependent variable

Question 12

What equation do we use ot indicate effect size?

Answer 12

cohen’s d = M1-M2 /SD pooled

Question 13

What are IV’s sometimes called when looking at correlation

Answer 13

No longer working with group means, but instead interested in correlation

When working with continuous independent variables, IVs are sometimes called ’predictors’

Question 14

What is the effect size for correlation? What equation do we use?

Answer 14

How strong is the relationship between variables?

Example: Health and Life Satisfaction

Need a metric we can compare across different studies

Effect size = Strength of relationship / Variability

“Covariance” or shared variance
Pearson product-moment correlation, or

Magnitude of the relationship between the predictor and the dependent variable

r = covariance (x,y) / S.D. (x) S.D. (y)

Question 15

What is the point of regression?

Answer 15

Using correlations between variables to make predictions

Question 16

What should always be the first step in analysis?

Answer 16

visualizing data

Question 17

LOOK AT SLIDE 22