Free online JSON to SAV converter

What the JSON format is

JSON is a UTF-8 text format built from objects, arrays, strings, numbers, booleans, and null values. Web applications, REST APIs, data pipelines, NoSQL databases, and database export tools commonly create JSON files; a typical dataset is an array of objects such as one object per respondent or transaction.

JSON does not require every object to have the same fields, and fields can contain nested objects or arrays. That flexibility is useful for interchange but does not directly match the rectangular rows-and-columns model expected by statistical software.

What the SAV format is

SAV is the IBM SPSS Statistics system data format. It stores rectangular records together with metadata such as variable names, variable labels, value labels, measurement levels, missing-value definitions, formats, and character encodings.

SPSS, PSPP, R packages such as haven, and Python libraries based on ReadStat can read SAV files. SAV is suitable when the recipient needs SPSS-compatible metadata or statistical analysis rather than a human-readable text export; it is not a general-purpose relational database format.

How to actually convert JSON to SAV

Convert with Python

Python is the most controllable option when the JSON contains arrays, nested fields, nulls, dates, or values that need type cleanup. Install the required packages:

python -m pip install pandas pyreadstat

Save the following as json_to_sav.py. It expects either a top-level JSON array or an object containing a records array.

import json
import pandas as pd
import pyreadstat

with open("input.json", encoding="utf-8") as file:
    data = json.load(file)

if isinstance(data, dict) and isinstance(data.get("records"), list):
    data = data["records"]

if not isinstance(data, list) or not all(isinstance(row, dict) for row in data):
    raise ValueError("Expected a JSON array of objects or an object with a records array")

df = pd.json_normalize(data, sep="_")
pyreadstat.write_sav(df, "output.sav")

Run it with:

python json_to_sav.py

pd.json_normalize turns nested paths into column names such as address_city. If the JSON uses another wrapper key, replace records with that key. If it contains a single object rather than records, identify which properties represent rows before converting.

Convert without installing software

You can convert the file for free on the 101convert.com website by uploading the JSON, choosing SAV as the output format, and downloading the generated file. Inspect the result in SPSS or PSPP, especially when the source contains nested data or mixed types.

Quality and compatibility considerations

• Flattening is a data decision: nested objects become separate columns, while arrays require a rule such as one row per array item, joined text, or separate related tables. A converter cannot infer the correct relational structure in every case.
• Mixed JSON types need cleanup: a column containing numbers, strings, and nulls may become a character variable or lose values during type inference. Convert each column deliberately in pandas before writing the SAV file.
• Dates are not automatically statistical dates: parse ISO date strings and assign the intended SPSS date or datetime representation if date calculations are required. Otherwise they remain ordinary character columns.
• Metadata is usually absent from JSON: variable labels, value labels, measurement levels, and user-missing values must be supplied separately. The conversion preserves values more reliably than it can reconstruct statistical meaning.
• Check names and lengths: use simple, unique column names compatible with the target SPSS version. Long names, punctuation, duplicate keys, and names beginning with unsupported characters may be changed or rejected.
• Validate the output: reopen output.sav with SPSS, PSPP, or pyreadstat.read_sav, then compare row counts, column counts, null handling, date values, and representative records with the source JSON.