Describe It
Uncertain Times
T for Two
Related Matters
Variance Vault
Excel LIVE!
100

The standard deviation is the square root of this measure of variability. 

Variance. 

100

This term describes doubt about statistical outcomes due to random error or chance.

Inferential uncertainty

100

Unlike a z test, a t test can be used when this value is unknown.

population standard deviation.

100

Pearson's r can range from this value to this value

−1 to +1

100

ANOVA is used instead of an independent t-test when comparing this many groups.

Three or more

100

On Sheet 1, find the mean, median and mode of the scores 

=AVERAGE(A2:A31) → 22.4

=MEDIAN(A2:A31) → 15.5 

=MODE(A2:A31) → 41 and 8

200

Winners of a race can be an example of this level of measurement when we don't know the gap between first and second place.

Ordinal. 

200

A sample will never perfectly mirror its population; the remedy for this type of random error is a larger random sample.

Sampling error. 

200

This is the standard alpha level in psychology

α= .05

200

On a 1–5 scale, a response of 2 becomes this value when reverse scored

4

200

Between-groups variance is attributed to this, while within-groups variance reflects individual differences and measurement error

the effect of the treatment (the IV)

200

On Sheet 1, produce a frequency table and histogram

Frequency via pivot table 

Histogram of the frequencies 

300

Temperature and IQ scores fall at this level of measurement, which has equal intervals but no true zero.

Interval. 

300

This measure answers the question, "How much would the sample average vary if we repeated this experiment many times?"

Standard error. 

300

This t-test compares two repeated measures from the same participants and uses df = n − 1.

The paired-samples t-test

300

Alcohol use and aggression are correlated, but a person's upbringing may drive both. This is the name for that kind of explanation

Third or extraneous variable

300

The F ratio is calculated by dividing these two terms

MSB divided by MSW (between-groups over within-groups mean squares)

300

On Sheet 4, find the mean, sample standard deviation and the upper and lower outliers for post scores. 

=AVERAGE(A2:A9) → 17.5

=STDEV.S(A2:A9) → 2.98

=µ ± SD*3 → 26.42 (upper) 8.57 (lower)

400

When a distribution is right (positively) skewed, the mean is pulled too high, so this is the better measure of central tendency.

Median. 

400

Our results tell us that p = .03 so we reject the null hypothesis, and this is what we can determine about our outcome.

There is a 3% change the outcome is due to random error/chance

P value – what is the probability a random process could produce a result as extreme as we got

400

Failing to reject a null hypothesis that is actually false is this type of error, which is related to statistical power.

Type II error

400

Correlation does not mean causation

Correlational fallacy

400

With 3 groups and 15 participants in total, these are the between-groups and within-groups degrees of freedom

2 and 12

400

On Sheet 2, find Pearson's r, then t and p for the correlation (n = 8).

=PEARSON(A2:A9,B2:B9) → r = .978; 

t = =r*SQRT((8-2)/(1-r^2)) → 11.40;

 p = =T.DIST.2T(ABS(t),6) → .00003 (p < .001)

500

Using the empirical rule, about this percentage of scores fall within 2 SD of the mean, and a score beyond this many SDs is considered an outlier.

About 95.4% and 3 SD

500

Nearly every test statistic in this unit, including z, t, r, and F, has this same two-part structure. Name both parts and say what each one represents

information divided by error. 

The information term is the evidence that something real is going on (e.g., the difference between means for t). The error term estimates the random error in the data (e.g., within-group variance for F). A larger ratio makes a chance result less plausible

500

For t(166) = 16.77, p < .001, d = 1.29, this is the decision about the null and the size of the effect

reject the null, with a large effect (d ≥ .80)

500

How much the scores vary in the same way when correlating them

Covariance

500

What does this tell us about the data and how does the F-stat explain the outcome

Numerator: Measures how different group means are from the grand mean

Denominator: Measures how much each score varies around its own group mean

Larger values of F indicate larger treatment effects

500

On Sheet 3, run the correct ANOVA and report the group means, F, and p. Interpret the results.

Means 8, 5, 3; SSbetween = 63.33, SSwithin = 30.00 (df = 2 and 12); MSbetween = 31.67, MSwithin = 2.50; F = 12.67, p = .0011. Significant; reject the null