data-analysis

name	data-analysis
version	1.0.0
description	Guide for quantitative, qualitative, and mixed methods data analysis. Statistical test selection, interpretation, and tool recommendations (SPSS, R, Python).
allowed-tools	["Read","Write","Edit"]

Data Analysis Helper

Guide untuk data analysis — quantitative, qualitative, dan mixed methods.

Quick Navigation

• Quantitative → Descriptive, Inferential stats, SPSS/R/Python
• Qualitative → Thematic analysis, coding, NVivo/ATLAS.ti
• Mixed Methods → Integration strategies
• Tool Selection → Which software for your needs?

---

QUANTITATIVE ANALYSIS

Step 1: Data Preparation

Checklist:

• Clean data (handle missing values, outliers)
• Check variable types (nominal, ordinal, interval, ratio)
• Verify sample size adequate
• Check normality (Shapiro-Wilk, Kolmogorov-Smirnov)

Step 2: Descriptive Statistics

Statistic	When to Use	Interpretation
Mean	Normally distributed continuous data	Average value
Median	Skewed data, outliers present	Middle value
Mode	Categorical data	Most frequent
SD	Variability of continuous data	Spread around mean
Range	Quick variability check	Min to max
Percentage	Proportions	Part of whole

Reporting Format:

M = 25.4, SD = 3.2 (for normally distributed)
Mdn = 24, IQR = 18-28 (for skewed data)

Step 3: Inferential Statistics — Test Selection

Flowchart:

1. How many groups?
   ├── One group → One-sample t-test / Chi-square
   ├── Two groups → 
   │   ├── Independent → Independent t-test / Mann-Whitney U
   │   └── Related/Paired → Paired t-test / Wilcoxon
   └── Three+ groups →
       ├── Independent → ANOVA / Kruskal-Wallis
       └── Related → Repeated measures ANOVA / Friedman

2. Variable relationships?
   ├── Two continuous → Pearson/Spearman correlation
   ├── Predict outcome → Regression (Linear/Logistic)
   └── Multiple predictors → Multiple regression

Test Selection Table:

Research Question	Variable Types	Test
Compare group means	1 categorical (2 groups), 1 continuous	Independent t-test
Compare before-after	2 related continuous	Paired t-test
Compare 3+ groups	1 categorical (3+ groups), 1 continuous	One-way ANOVA
Association between variables	2 categorical	Chi-square
Relationship strength	2 continuous	Pearson/Spearman
Predict outcome	1+ predictors, 1 outcome	Regression

Step 4: Assumptions Check

Test	Assumptions	If Violated
t-test	Normality, homogeneity of variance	Use Mann-Whitney U
ANOVA	Normality, homogeneity, independence	Use Kruskal-Wallis
Pearson	Linearity, normality, homoscedasticity	Use Spearman
Regression	Normality, linearity, no multicollinearity	Transform data

Step 5: Effect Size

Test	Effect Size	Small	Medium	Large
t-test	Cohen's d	0.2	0.5	0.8
ANOVA	η² (eta squared)	0.01	0.06	0.14
Correlation	r	0.1	0.3	0.5
Chi-square	Cramér's V	0.1	0.3	0.5

Always report effect size! Not just p-value.

Step 6: Reporting Results

Template:

A [test] revealed [significant/non-significant] differences/effects/relationships
[description of finding], [df] = [value], p = [value], [effect size] = [value].

Example:
An independent samples t-test revealed a significant difference in test scores
between Group A (M = 78.5, SD = 5.2) and Group B (M = 72.1, SD = 6.8),
t(48) = 3.45, p = .001, d = 1.05, 95% CI [0.42, 1.68].

---

QUALITATIVE ANALYSIS

Step 1: Familiarization

• Read through all data multiple times
• Note initial impressions
• Keep reflexive journal

Step 2: Coding

Types of Coding:

• Open coding → Initial categories from data
• Axial coding → Link categories to subcategories
• Selective coding → Core category identification

Software:

• NVivo (comprehensive, expensive)
• ATLAS.ti (flexible)
• MAXQDA (user-friendly)
• Dedoose (cloud-based, affordable)

Step 3: Thematic Analysis

Braun & Clarke (2006) Six Steps:

1. Familiarize with data
2. Generate initial codes
3. Search for themes
4. Review themes
5. Define and name themes
6. Write up

Step 4: Quality Criteria

Criterion	Description
Credibility	Believable results; member checking
Transferability	Context-rich description
Dependability	Audit trail, consistent process
Confirmability	Researcher reflexivity

Step 5: Reporting

Template:

Data were analyzed using thematic analysis following Braun and Clarke's 
(2006) six-step approach. Three main themes emerged: [theme 1], [theme 2], 
and [theme 3]. These themes were developed through iterative coding of [N] 
[interviews/focus groups/documents].

---

MIXED METHODS

Integration Strategies

Strategy	When to Use
Merging	Compare quantitative + qualitative results side-by-side
Explanation	Qual explains quant findings
Exploration	Qual explores first, quant follows up
Embedding	One method supports the other

Design Types

• Convergent Parallel → Collect both simultaneously
• Explanatory Sequential → Quant first, then qual
• Exploratory Sequential → Qual first, then quant

---

TOOL SELECTION GUIDE

Tool	Best For	Pros	Cons
SPSS	Social sciences, beginners	GUI, widely accepted	Expensive, limited customization
R	Advanced stats, reproducibility	Free, powerful, flexible	Steep learning curve
Python	Large datasets, ML integration	Flexible, industry standard	Requires coding
JASP	Open-source alternative to SPSS	Free, user-friendly	Less features than SPSS
NVivo	Qualitative analysis	Comprehensive	Expensive
ATLAS.ti	Text/multimedia analysis	Flexible	Expensive

---

COMMON MISTAKES

• ❌ Reporting p-values without effect sizes
• ❌ Using parametric tests on non-normal data
• ❌ Ignoring assumptions checking
• ❌ Confusing correlation with causation
• ❌ Over-interpreting qualitative themes
• ❌ Not reporting software version

---

PROCESS

1. Identify research questions → What do you want to know?
2. Check data type → Quant/Qual/Mixed?
3. Select appropriate tests → See tables above
4. Verify assumptions → Critical before analysis
5. Run analysis → Document everything
6. Interpret results → Statistical + practical significance
7. Report findings → Use templates above
8. Visualize → Tables, figures, charts

---

QUICK REFERENCE COMMANDS

SPSS

DESCRIPTIVES VARIABLES=score /STATISTICS=MEAN STDDEV.
T-TEST GROUPS=group(1 2) /VARIABLES=score.
ONEWAY score BY group /STATISTICS DESCRIPTIVES.
REGRESSION /DEPENDENT outcome /METHOD=ENTER predictor1 predictor2.

R

# Descriptive
describe(data)
summary(data)

# t-test
t.test(group1, group2)

# ANOVA
aov(outcome ~ group, data)

# Correlation
cor.test(x, y)

# Regression
lm(outcome ~ predictor, data)

Python

import pandas as pd
from scipy import stats

# t-test
stats.ttest_ind(group1, group2)

# ANOVA
stats.f_oneway(group1, group2, group3)

# Correlation
stats.pearsonr(x, y)

# Regression
from sklearn.linear_model import LinearRegression