File formats
Open a Triple-S survey file
Select the Triple-S metadata (.sss or .xml) and its data file (.asc, .dat or .csv) together in crosstabs.com. Both are read in your browser with variable labels, value labels and multiple-response questions taken from the metadata, ready for crosstabs with significance testing. Nothing is uploaded.
How results are checked
Named statistical review pending. Selected automated reference cases are published on the validation page.Run this on your own data — free, no signup
Upload a CSV, Excel, SPSS, Stata, SAS, R, Parquet or JSON file. Everything runs in your browser; your file never leaves your device.
Open the workspace →Why Triple-S comes as two files
Survey and tabulation software exchange data as Triple-S because the data file alone is only columns of codes: the metadata says where each variable sits in a record, what its codes mean, and which questions allow several answers. Crosstabs reads both, so a single question shows its answer labels and a multiple question becomes one table row per category.
What you can do with it
Everything the workspace does for SPSS, CSV and Excel works for Triple-S data: crosstabs with chi-square and column significance letters, multi-select tables, weighting, banner tables and CSV, Excel or SPSS .sav export. The import formats guide lists the limits that apply to every import.
Using it well
How to interpret it
Rule of thumb
The metadata is the point of Triple-S: keep both files together and the labels and multiple-response questions arrive with the data.
Questions
Frequently asked questions
Which files do I need to select?
Both files of the dataset, together: the XML metadata (.sss or .xml) and the data file it describes (.asc or .dat for fixed-width records, or .csv). Select or drop them in one go. If only one of the two is chosen, Crosstabs says which one is missing, and names the data file when the metadata records it.
Which Triple-S versions and variable types are read?
Triple-S 1.2, 2.0 and 3.0 surveys with one record per respondent, in fixed-width or CSV format. All seven variable types are read: single (numeric or literal codes), multiple (spread and bitstring layouts), quantity (with decimals and negative values), character, logical, date (YYYYMMDD) and time (HHMMSS).
What happens to multiple-response questions?
Each multiple variable becomes one 1/0 column per category (for example Q2_1, Q2_2) and a multi-select question built from the metadata, so tables show each category as a row with % of cases. Nothing is guessed from column names.
How are missing values and filters handled?
As the standard says: a field of spaces (or an empty CSV field) is missing, and unused spread subfields are ignored. A variable with a <filter> is missing for respondents whose filter variable is not true. Values that do not match the declared type are read as missing and counted in the import notes.
Is anything uploaded?
No. Both files are read in your browser. The XML reader never fetches a DTD or other external resource and refuses entity declarations.
What is not supported?
Hierarchical Triple-S datasets (several linked record levels) and files with more than one record per respondent are refused with an explanation; export the respondent level as its own survey. Interview-only text, routing and score-based statistics are not used beyond keeping scores in the variable notes. A weight variable is marked as a weight but is only applied when you choose it.
Sources and next steps
References & further reading
Try it on your own data — free, no signup
Upload a CSV, Excel, SPSS, Stata, SAS, R, Parquet or JSON file. Everything runs in your browser; your file never leaves your device.
Open the workspace →