How can we help?

You can also find more resources in our Help Center.

Construct Validity

Is my IV manipulating what I want it to manipulate?

Is my DV measuring what I want it to measure?

Is my DV measuring what I want it to measure?

Internal Validity

Is my experiment a fair test of my hypothesis?

External Validity

Do my findings generalise to other populations, or other variables?

Reject Null when Null is true

Type 1 error

Accept Null when Null is not true

Type 2 error

Null Hypothesis

H0 = There is no effect of the IV on the DV

e.g. Music has no effect on math

e.g. Music has no effect on math

Research Hypothesis

H1 = There is an effect of the IV on the DV

e.g. Music has an effect on math

e.g. Music has an effect on math

t = 0

No variability between groups (they are drawn from same population)

p-value

The likelihood that you obtain the observed result (or a result more extreme), given the null hypothesis is true

Two-tailed p-value

0.25 at each end of the distribution curve

One-tailed p-value

0.5 at the predicted direction of the distribution curve

Effect sizes

Measure of variability due to my effect divided by variability in my sample. P-value says nothing this. Different statistics have different measures of it.

Cohen's d

The bigger the difference between means, the bigger the effect size. The bigger the SD is, the smaller the effect size

Within-subjects design

Advantages: fewer subjects, more statistical power

Disadvantages: longer experiments, counterbalancing, carryover effects

Disadvantages: longer experiments, counterbalancing, carryover effects

Stratified Random Sampling

More specific populations e.g. gender, culture, handedness

Oneway ANOVA

More than 2 groups to be compared.

Between-group variance

F = -------------------

Within-group variance

Between-group variance

F = -------------------

Within-group variance

Post-hoc tests

Comparisons of means after finding a significant F.

Used when I have no hypothesis about how the means might differ from each other (2-tailed).

Used when I have no hypothesis about how the means might differ from each other (2-tailed).

df(x, y)

X = # of groups - 1

Y = total # - # of groups

Y = total # - # of groups

Multiple Comparisons

When you have 4 groups, and you only want to compare 2 of them, use an independent t-test