{
  "cells": [
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "# Games–Howell in Python with statsmodels 0.15\n",
        "A reproducible notebook for the Salar Cafe step-by-step guide.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 1. Setup\n",
        "Install packages with `python -m pip install -r requirements.txt` and place the CSV beside this notebook.\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "import math\n",
        "from pathlib import Path\n",
        "import numpy as np\n",
        "import pandas as pd\n",
        "import matplotlib.pyplot as plt\n",
        "from scipy import stats\n",
        "from statsmodels import __version__ as statsmodels_version\n",
        "from statsmodels.stats.oneway import anova_oneway\n",
        "from statsmodels.stats.multicomp import pairwise_tukeyhsd\n",
        "print(statsmodels_version)\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 2. Load and validate the data\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "order = ['Control','Video','Interactive','Tutoring']\n",
        "df = pd.read_csv('posthoc_training_scores.csv')\n",
        "df['group'] = pd.Categorical(df['group'], categories=order, ordered=True)\n",
        "display(df.head())\n",
        "display(df.groupby('group', observed=True)['score'].agg(['count','mean','std','median','min','max']))\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 3. Variance diagnostic\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "arrays = [df.loc[df['group']==g, 'score'].to_numpy() for g in order]\n",
        "stats.levene(*arrays, center='median')\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 4. Standard and Welch ANOVA\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "classic = anova_oneway(df['score'], df['group'], use_var='equal')\n",
        "welch = anova_oneway(df['score'], df['group'], use_var='unequal')\n",
        "classic, welch\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 5. Tukey HSD and Games–Howell\n"
      ]
    },
    {
      "cell_type": "code",
      "execution_count": null,
      "metadata": {},
      "outputs": [],
      "source": [
        "tukey = pairwise_tukeyhsd(df['score'], df['group'], alpha=.05, use_var='equal')\n",
        "games_howell = pairwise_tukeyhsd(df['score'], df['group'], alpha=.05, use_var='unequal')\n",
        "display(tukey.summary_frame())\n",
        "display(games_howell.summary_frame())\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 6. Read the decision reversals\n",
        "In the supplied data, Control vs Video changes from Tukey non-significant to Games–Howell significant, while Video vs Interactive changes in the opposite direction.\n"
      ]
    },
    {
      "cell_type": "markdown",
      "metadata": {},
      "source": [
        "## 7. Continue with the full script\n",
        "The bundled `.py` file adds effect sizes, transparent Games–Howell verification, CSV exports, and all four figures used in the article.\n"
      ]
    }
  ],
  "metadata": {
    "kernelspec": {
      "display_name": "Python 3",
      "language": "python",
      "name": "python3"
    },
    "language_info": {
      "name": "python",
      "version": "3"
    }
  },
  "nbformat": 4,
  "nbformat_minor": 5
}