Cheatography
https://cheatography.com
3 page sheet + matrix for exam (double sided)
Types of Data
Categorical/Nominal |
Do not hold numerical meaning (arbitrary) |
Ordinal |
Rank ordering, differences not equal |
Interval |
Intervals between points on a scale are equal and the same, zero is arbitrary |
Ratio |
Zero is NOT arbitrary (an absence) |
Experimental Designs
Balanced |
each cell (each combination of factors) contain the same number of replications (how many measurements |
Complete |
every level of one factor combined with every level of the other factor(s) |
Incomplete |
Lots of factors or many measurements (nested/block design best) |
Single subject/repeated measures |
Subject acts as their own control |
Ceiling effects: Test is too easy (100%)
Floor effects: Test is too hard (0%)
Learning effects: subjects improve with more trials
Order effects: test order may have effect on outcome
Characteristics of Data Sets
Data Shape |
Frequency distributions are a common way to describe data shape (range of scores) |
Location |
finding central tendency or middle of data |
Spread |
Variance -> range, SD and IQR |
Outliers |
Clustering |
e.g. bimodal distribution |
Granularity |
Data only takes on certain values (e.g. discrete data + rounded continuous) |
|
(e.g. discrete data + rounded continuous) |
Types of Sampling
Random |
Increased ability to generalise to population |
Systematic |
Choosing subjects from a population at a regular interval (choosing every second item) |
Cluster |
Randomly select a few schools in your sample and have all students as participants |
Convenience |
Sample used because it is accessible rather than representative of a population |
|
|
Central Limit Theorem
• draw a large enough sample from the population and plot all of those sample means, our sampling distribution will approach normal
• Sampling distribution uses sample means
• Population mean: mean of all sample means
Standard Error
- SD of sampling distribution
95% CI = sample mean +- 1.96 x SE |
Pearson's Correlation (r)
Strength |
Positive |
Negative |
Strong |
.8 to 1 |
-.8 to -1 |
Moderate |
.5 to .7 |
-.5 to -.7 |
Weak |
0 to .4 |
0 to -.4 |
ANOVA Variance
|
DF |
Sum of Squares |
Mean Sqaure |
Between Groups |
no. groups -1 |
How much data varies between different groups (variance) |
Average variance between groups |
Within Groups |
no. data points - no. of groups |
How much data varies within each group (variance) |
Average variance within groups |
Total |
no. data points - 1 |
Types of ANOVAs
One-way |
1 factor/independent variable (categorical) |
Two-way |
2+ factors/IVs (categorical), interactions |
Repeated Measures |
Measure the same outcome variable on the same population twice |
|
Each subject is now a random factor (rather than fixed factors) |
T-test Types
Test |
Description |
DF |
1-sample (single) |
Compares your experimental group with a hypothesised or known value |
n-1 |
2-sample (independent) |
Compares the means for two independent samples |
(n1-1) + (n2-1) |
Paired |
measuring something for the same group of people |
n-1 |
One tailed: Directionless -> one group is different from the other group (in pos or neg direction)
Two tailed: Directional -> one group if larger or smaller than the other
Linear Regression
Beta |
degree of change in the outcome variable for every 1 unit of change in the predictor variable |
R-Sqaured |
Fit of the model and represents how much variance in the DV can be accounted for by the IV |
Analysis of Variance |
Adj SS (adjusted sum of squares) -> total variance of data |
|
- The error SS is what is left over -> variance that cannot be explained by other factors or variance in the model |
Predicting
CI: If we repeated our experiment many times an degenerated a confidence interval each time, 95% of those confidence interval will contain the true population value
Prediction Interval: Predicting future observations from the regression equation
|
|
Assumptions of Parametric Tests
1. Normally distributed data
2. Homogeneity of variance
3. Interval/ratio data
4. Independence
This means that you may have to use non-parametric tests when...
• your data is better represented by the median (e.g. skewed data like salary or house prices), or
• you have a very small sample size, or
• you have ordinal (e.g. rating scales, some questionnaire results) or categorical data |
Parametric and Non-Parametric Equivalent Tests
Parametric |
Non-Parametric |
1-sample AND paired t-test |
Sign test or Wilcoxon signed-rank test |
2-sample t-test |
Mann-Whitney test |
One-way ANOVA |
Kruskal-Wallis test |
Multifactor ANOVA (two-way + repeated measures) |
N/A |
N/A |
Chi-square test |
Types of Qualitative Data
Transcripts (e.g. interview) |
Allows the researcher to ask about specific things and probe deeply |
Observation |
ethnographic studies |
Pictures |
Pictures could be photos that the researcher has taken (drawings, rooms etc) |
Documents |
Many types (e.g. progress notes) |
Web content |
Publicly available (e.g. social media) |
Sampling for a Qualitative Study
Typical Case |
Average case |
Extreme case |
Unusual, unique or distinct case |
Maximum Variation |
Looking for the biggest range of perspectives |
Homogenous Group |
Minimum variation sampling + Focus on in-depth area of interest |
Stratified Purpose |
Selected cases from identified subgroups (e.g. 5 people from 4 age groups) |
Theoretical |
Start data collection -> analyse results -> form therapy -> continue sampling |
Snow Ball |
One respondent is asked to suggest others. |
Convinience |
Recruiting anyone who is at hand |
Qualitative Evidence
Tangibly (concrete) |
Intangibly |
Guidelines, protocols |
understanding what clients want from their clinicians |
practice recommendations based on qual research |
broaden knowledge and change behaviours |
Setting up a Qualitative Analysis
Deductive (top-down) |
Inductive (bottom-up) |
coding will be influenced by the framework you're using |
coding will be purely based on what the participant has said, without trying to fit it into a framework. |
|
Created By
Metadata
Comments
No comments yet. Add yours below!
Add a Comment
Related Cheat Sheets