Research and Statistical Designs

Question 1

Confidence Intervals

Answer 1

Measure of the precision of the inferential statistic-based estimate of the true population value

Typically set at 95%

e.g. mean: based on the sample data we can be 95% confident that the interval contains the population mean

Question 2

Doing a Meta-Analysis

Answer 2

1. Choose research question
2. Choose studies
3. Calculate summary effect size (mean) & heterogeneity (variance)
4. Check for publication bias
5. Regression analysis to check moderators (did some variables influence the effect size?)

Question 3

Asymmetry Test

Answer 3

Test for publication bias

You should get a symmetrical funnel plot: shows that effect sizes either side of significance threshold were published

Question 4

MaxMinCon

Answer 4

Maximise experimental variance

Minimise error variance (sampling & measurement error)

Control extraneous variance

Question 5

Sampling Error

Answer 5

Difference between sample mean and population mean

Reduced by random sampling (experimental research)

Question 6

Quasi-experimental research

Answer 6

Like experimental (i.e. looking for causal effect)

No random assignment, but has a control group or multiple measures

Question 7

Measurement Error

Answer 7

Difference between observed and true score

Random vs systematic

Can never be eliminated

Question 8

Measurement Error

Answer 8

Difference between observed and true score

Random vs systematic

Can never be eliminated

Question 9

THIS MESS DREAD

Answer 9

Threats to internal validity

THIS MESS: can be ruled out during study design
DREAD: can be ruled out during study conduction

Question 10

THIS MESS

Answer 10

Testing effect
History
Instrumentation
Selection

Maturation
Experimental mortality
Statistical regression
Selection-maturation interaction

Question 11

DREAD

Answer 11

Diffusion of experimental effect
Rivalry
Equalisation of treatments
Ambiguous temporal precedence
Demoralisation

Question 12

Statistical Tests

Answer 12

T-tests (standard or paired)
ANOVA (one-way, RM or multifactor)
ANCOVA (one-way)
Mann-Whitney U test
Regression analysis (linear or multiple)
Pearson’s coefficient of correlation (r)

Question 13

T-test

Answer 13

For comparing means of two groups

Standard (aka independent) T-test:
post-test only, between-subjects design

Paired (aka repeated measures) T-test:
pre-post, within-subjects design

Question 14

ANOVA

Answer 14

For comparing means of 3+ groups

One-way ANOVA:
post-test only, between-subjects

Simple repeated measures ANOVA:
pre-post, within-subjects

Mulifactor (aka factorial; e.g. two-way) ANOVA:
multiple levels for each group, between-subjects or mixed design
[Factor = IV]

Question 15

One-Way ANCOVA

Answer 15

Mix of ANOVA and regression

For comparing means of 3+ groups, whilst controlling for scale covariate(s)

Way of eliminating within-group error variance on the dependent variable –> increased precision, increased statistical power

Question 16

Mann-Whitney U Test

Answer 16

Non-parametric equivalent to standard t-test
(i.e. post-test only, between-groups)

Discrete, ordinal DV –> converted to ranks

Question 17

Regression Analysis

Answer 17

For measuring the extent to which IV(s) relate to a DV

Linear Regression:
one IV

Multiple (aka multivariate) Regression:
multiple IVs

Question 18

Pearson’s Coefficient of Correlation (r)

Answer 18

Measure of correlation between two variables