Method guide / Configurations

Expanded fsQCA

Run a larger configurational study while keeping search size, thresholds and solution alternatives explicit.

16 min read

What the method does

As conditions are added, the possible configuration space grows rapidly and more combinations may remain unobserved. Expanded fsQCA organises the supported larger analysis, but its methodological questions remain calibration, case coverage, necessity, sufficiency and counterfactual assumptions. More conditions should follow the theory rather than a search for a preferred result.

k binary conditions imply 2ᵏ possible configurations.

Adding one binary condition doubles the space. The number of observed configurations may be much smaller, increasing the importance of sparse rows and logical remainders. Examine the completed search status as well as any formulas returned.

Data and analysis in QuickPLS

Use a justified condition set, calibrated memberships and a clear case population. Review the resulting configuration count and thresholds before calculation.

QuickPLS provides the expanded-study controls and result tables. Check completion, alternatives and the contributing cases before reporting a solution.

Follow a worked example

The following example shows the data, settings and decisions for one analysis. Use the linked practice files to follow it. For your own study, keep the procedure and choose the variables, model and settings that fit your research question.

Download memberships.csv to begin. It contains 10 rows and 5 columns. The example below identifies the variables to use, their roles and any rows to exclude; not every column is an analysis variable.

More conditions can make even a small dataset demanding

A dataset's row count is only part of the calculation. Each binary condition doubles the possible combinations, called configurations. Sixteen conditions create 65,536 configurations. The work also depends on which combinations are positive, which are unobserved and how many different complete solutions the table allows.

Expanded fsQCA is a separate workspace that you select deliberately. Learn it with a small saved study whose cases and set definitions you already understand. Load the setup, estimate the resources and inspect the truth table before scaling up. The estimate belongs to those data, calibration choices and cases; changing them may require another estimate.

Use a familiar study as the starting point

Begin with a small saved study you can explain and check. Read the expanded-mode acknowledgement before enabling it. Keep enough free storage for the protected project copy, and note where that copy is saved so you know which project contains the expanded work.

Estimate the work before calculating

  1. Load the saved setup

    Choose Data, More, Expanded fsQCA. Select the owning study and saved revision, then click Load expanded fsQCA setup. Review Enable expanded mode with bounded resources.

  2. Read the resource estimate

    Click Estimate expanded resources. Inspect the configuration count and stated limits before deciding to proceed.

  3. Check the truth table

    Click Preview expanded truth table. Confirm the selected conditions, outcome, included cases and configuration classifications.

  4. Save the expanded revision

    Click Save expanded fsQCA revision. Follow the activation step and use the protected copy. Reload the saved setup or estimate again if the app requests it.

  5. Calculate and inspect the result

    Select the intended source dataset and click Calculate expanded fsQCA. Read the completion status before examining returned whole-solution alternatives, their case pages and the matching report.

Treat the stated sizes as conditional limits

Expanded mode targets up to 16 conditions or 50,000 original cases within fixed calculation limits. It does not promise that both maxima can be used together or that every smaller setup will finish. The calculation allowance controls the method's work; it does not measure all memory used by the application. Data preparation, saved output size and indirect-calibration assignments also have their own limits.

If the app rejects a setup or warns about a limit, read which part of the work caused it. Check for a wrong dataset or accidentally selected conditions before changing the study. Removing intended cases or conditions changes the research scope and needs an explanation. A cancelled or resource-stopped attempt remains incomplete. For a completed result, inspect an actual returned alternative and its cases, then use the report supported by that result type.

The estimate describes the selected setup, not a guarantee of completion. Expanded results use their own workflow rather than ordinary robustness or multiple-outcome plans, and a report can only include appendices or pairings supported for that type. Case-page navigation changes which rows you see; it does not reduce the calculation's included cases.

On a small screen, swipe the table sideways to see every column.

OutputHow to read itWhat to check
Resource estimateRead the estimate for the selected dataset, calibration, cases and conditions.Changing the specification may require a new estimate. An estimate does not promise a completion time.
Configuration and search limitsCheck how many configurations are possible, how much simplification is needed and how many alternatives may be returned.Sixteen conditions or 50,000 cases are separate supported sizes. They do not guarantee that both maxima can be combined.
Completion statusKeep a record if a limit was reached, the calculation was cancelled or the setup was rejected.A partial search does not provide a complete family of solutions.
Selected alternative and case pagesInspect the pages for a chosen saved alternative and its matching expanded report.The number of rows displayed per page only changes how results are shown. Expanded results cannot be added to ordinary robustness or outcome plans.
Screenshot to be addedExpanded fsQCA solution with its selected pathway and original case details

Learn the controls with a checkable small project

Save a working copy of Guided fsQCA. Keep its supplied ordinary teaching result as a reference while you learn the separate expanded workflow. Compare the two only after your own expanded calculation finishes.

Start with the delivery example

Use memberships.csv and its saved guided study. Read A as capacity, B as reach, C as coordination and Y as delivery. The stated crossover and incomplete-case exclusions leave m01 to m08. A crossover is a membership exactly on the boundary between mostly in and mostly out. Keeping these case decisions explicit makes the small result checkable before you explore a larger setup.

Expanded fsQCA targets larger studies, including up to 16 conditions or 50,000 original cases, subject to resource checks and other limits. This lesson uses the small example to teach preparation and planning. It does not present a newly completed large-data run, an accuracy comparison or a speed benchmark.

Understand why the planned study needs more work

Adding cases means more observations to prepare and evaluate. Adding binary conditions doubles the possible configurations each time, which is exponential growth. It can also leave many configurations unobserved and allow many different ways to simplify the truth table. These demands are different even when the CSV file looks small.

Keep a research reason for every condition. For the teaching baseline, use existing memberships, frequency 1, consistency 0.80 and PRI 0.50, with no contradiction band. Frequency sets the minimum case support; consistency checks agreement with the outcome; PRI helps identify combinations that also fit its absence. Retain both stated exclusions. Keeping these choices fixed lets you learn the resource estimate without changing the research question.

Check the setup through to a completed result

  1. Load the delivery setup

    Open a copy of fsqca_guided_study.qpls and select memberships.csv. Choose Data > More > Expanded fsQCA. Select the delivery study and Existing memberships: explicit crossover and missing-case decisions revision, then Load expanded fsQCA setup.

    What you should see: The intended delivery setup is loaded.

    How to read it: Check its cases and condition definitions.

  2. Confirm the selected data and method

    Read the dataset, A/B/C conditions, Y outcome, existing memberships and exclusions. Do not select the neighbouring raw-score logistic example. Read and select Enable expanded mode with bounded resources.

    What you should see: The data and expanded-analysis settings correspond to the study.

    How to read it: Confirm the selected outcome direction.

  3. Estimate the calculation

    Click Estimate expanded resources. Read the configuration count and any warning or limit. Decide whether the intended setup can proceed before starting the calculation.

    What you should see: The estimated configuration space reflects the selected conditions.

    How to read it: Review unnecessary conditions before starting an expensive search.

  4. Preview the table

    Click Preview expanded truth table. For this exercise, check all eight configurations and the three positive rows against the small example below.

    What you should see: The preview truth table identifies observed rows and thresholds.

    How to read it: Inspect sparse rows and remainders.

  5. Save the expanded revision

    Click Save expanded fsQCA revision and complete the activation acknowledgement. Continue in the protected project copy and check its saved location.

    What you should see: The expanded revision is saved with its assumptions.

    How to read it: Retain the earlier baseline separately.

  6. Refresh after activation if requested

    If the app asks for a reload or a new estimate, select the actual saved source, load that setup and estimate again. An earlier estimate may not apply to the newly activated state.

    What you should see: The active setup reflects the saved revision.

    How to read it: Resolve any requested refresh before running.

  7. Run and check completion

    Click Calculate expanded fsQCA. Read whether the work completed, was cancelled or stopped at a resource limit. Treat only actual returned complete alternatives from a completed calculation as completed solutions.

    What you should see: The completed result states search status and available alternatives.

    How to read it: Interpret formulas only within the work that completed.

Screenshot to be addedthe exact setup

Count how quickly the possible combinations grow

Three binary conditions give 2 to the third power = 8 configurations. Twelve give 4,096; sixteen give 65,536. Moving from a 12-condition study to sixteen conditions multiplies the configurations by 16. It adds four conditions but far more than four truth-table rows.

In the small exercise, each of the eight configurations has one included case. A much larger condition set may have observations in only a small fraction of its possible configurations, even with many more cases. Those unobserved combinations are logical remainders. More calculation capacity does not supply the missing observations or justify assumptions about them; the frequency rules and simplifying assumptions still need research reasons.

On a small screen, swipe the table sideways to see every column.

Binary conditionsPossible configurationsRelative to three conditions
381 times
124,096512 times
1665,5368,192 times
Screenshot to be addedCheck the matching saved example

Use the ordinary saved result as a reference

The eight included cases have positive configurations 101, 110 and 111. Hand simplification gives A*B + A*C: capacity with reach, or capacity with coordination. The complete solution has consistency and coverage of 1. Each pathway has raw coverage 2/3 and unique coverage 1/3; raw coverage includes shared cases, while unique coverage counts only the contribution not covered by the other pathway. The ordinary saved membership result contains these values.

Use the numbers to check that the selected data and settings describe the intended study. They do not establish a newly executed expanded result, a speed measurement or agreement for every larger problem. After a completed run, compare its actual expression, fit and completion information with this small reference. A two-condition result image cannot verify this three-condition delivery exercise.

Read the estimate before deciding to proceed

The resource estimate checks the selected setup against the calculation limits. The 128 MiB allowance is a logical limit for the calculation kernel, the part that performs the analysis. It is not a measurement or cap for the whole app's memory use. Read it with the case and configuration counts. A small file can still require a difficult search when many complete solution alternatives are possible.

If work stops, first identify the stated reason. Check for an unintended dataset, extra selected columns or preparation of variables outside the research question. Fixing a mistaken selection can remove unnecessary work without changing the intended study. If the intended setup itself is too large, justify any narrower analysis and save it as a separately named revision.

Make a narrower study an explicit decision

A preparation limit, a large configuration space and too many alternative solutions are different problems. Read which limit was reached, then reconsider the research question or analysis design. If you remove intended cases or conditions, explain the reason and preserve the original plan. Do not keep reducing the data until a run finishes and report it as the unchanged study.

Indirect calibration retains its 10,000-assignment limit. Expanded source capacity does not enlarge every specialist preparation option. Ordinary robustness and multiple-outcome plans remain separate, and the expanded report does not automatically gain their appendices or a core/peripheral comparison. Plan the follow-up work using options supported by the actual result type.

Keep an interrupted attempt separate from a result

Cancellation or resource exhaustion means the requested calculation did not finish under the selected settings. A progress message or preview is not a complete set of solutions. Keep the original study and saved expanded revision so the attempted work and its choices remain clear.

After interruption, reopen the known protected project copy and inspect what was actually saved before restarting. A successful truth-table preview does not establish that simplification finished. If you decide to use a smaller problem, name and explain the new revision instead of replacing the account of the original attempt.

Identify the specific limit before changing the question

Check the saved source and included cases, read the estimate and identify the failure or stopping message. Use that information to distinguish a wrong selection from a resource problem in the intended analysis. Keep the message with the attempted revision so a later change has a clear reason.

On a small screen, swipe the table sideways to see every column.

SymptomWhat to checkAction
A below-limit study still exhausts resourcesThe candidate terms, alternative solutions and limit reported by the appIdentify which step reached its limit. Narrow the analysis only for an explicit research reason.
The estimate differs after savingWhether the saved expanded setup was reloadedLoad the returned specification and estimate again when requested.
A large source fails indirect preparationThe 10,000-assignment limitUse a supported, justified preparation approach rather than assuming expanded capacity removes that limit.
A cancelled calculation is treated as finalCompletion status and whether alternatives are completeReport the incomplete attempt separately and retry or revise deliberately.
The expanded method is absent from an ordinary planThe methods that the plan supportsUse the separate expanded workflow and report.
A reopened project shows only the old studyWhere the protected project copy and its revision were savedOpen the actual returned project copy and verify its contents.

Report the study scope and the completed work

State the selected data, conditions, outcome, calibration, exclusions and resource choices. Report whether calculation completed before presenting fit or alternatives. The wording below describes the proposed teaching exercise; after a real run, replace it with the actions and results you actually observed.

ItemWhat this lesson establishes
Teaching problem3 conditions; 8 included cases; 8 configurations
Baseline referenceA*B + A*C; consistency and coverage 1
Reference originHand counts and existing ordinary saved result
Expanded run in this tutorialNot newly executed
Actual run reportingSelected settings, estimate, completion status and returned alternatives

Example Methods wording for the proposed exercise

The planned expanded analysis uses the synthetic eight-case delivery baseline, with its conditions, existing memberships and analysis choices held fixed. Preparation includes a resource estimate and inspection of the eight-configuration truth table. The final report will identify whether calculation completed and which alternatives were returned.

Compare the reference only after your own completed run

After a completed calculation, compare the returned expression and fit with the small baseline, recording the selected settings and completion information. If a limit stopped the work, explain that outcome and any revised plan. Keep the original and revised attempts separately identifiable.

  • Calculate the configuration count before adding conditions.
  • Explain why more cases and more possible solutions create different types of work.
  • Find the actual protected project copy and its saved expanded revision.
  • Explain why the calculation allowance is separate from total app memory use.
  • Distinguish a successful preview or partial progress from completed alternatives.
  • Report only agreement or performance measured in an actual completed comparison.

Check the limits for the whole planned workflow

The targets are 16 conditions OR 50,000 cases, subject to the fixed limits. Their combination is not guaranteed. Ordinary robustness and outcome plans do not accept expanded results, and indirect calibration still has its 10,000-assignment cap. Check the preparation and reporting options you need before planning the larger run.

Calculation comparisons

Further reading

Related guides