Add data/pisa/schema.json

data/pisa/schema.json

@@ -0,0 +1,34 @@
+{
+  "dataset": "PISA 2018 (OECD)",
+  "source": "https://webfs.oecd.org/pisa2018/ (direct SPSS downloads available)",
+  "access": "FREE direct download",
+  "scale": "~600K students, ~900 item columns, 80+ countries",
+  "files": {
+    "SPSS_STU_COG.zip (456 MB)": {
+      "CNTSTUID": "Student ID",
+      "CNT": "Country",
+      "BOOKID": "Booklet",
+      "CM*Q*S": "Math items (scored 0/1/2, 7=not administered)",
+      "CR*Q*S": "Reading items",
+      "CS*Q*S": "Science items"
+    },
+    "SPSS_STU_QQQ.zip (478 MB)": {
+      "CNTSTUID": "Student ID (merge key)",
+      "CNT": "Country",
+      "ST004D01T": "Gender (1=Female, 2=Male)",
+      "ESCS": "SES index (continuous)",
+      "IMMIG": "Immigration (1=native,2=1st,3=2nd gen)",
+      "LANGN": "Language at home",
+      "PV1MATH-PV10MATH": "Plausible values"
+    }
+  },
+  "download_urls": {
+    "cognitive": "https://webfs.oecd.org/pisa2018/SPSS_STU_COG.zip",
+    "questionnaire": "https://webfs.oecd.org/pisa2018/SPSS_STU_QQQ.zip"
+  },
+  "python_code": "\nimport pyreadstat\n# Download and extract the zips first, then:\ndf_cog, meta = pyreadstat.read_sav('COG/CY07_MSU_STU_COG.sav')\ndf_qqq, meta = pyreadstat.read_sav('QQQ/CY07_MSU_STU_QQQ.sav')\ndf = df_qqq.merge(df_cog, on=['CNTSTUID','CNT'])\n# Now df has items + demographics per student\n",
+  "has_demographics": true,
+  "has_item_responses": true,
+  "has_selected_answer": false,
+  "has_question_text": "PARTIAL ~30-50% in separate OECD PDFs"
+}