STUDY designs and variables

Why do I need this step?

Every plot and every test in the next lessons works on the current design. If you pick the wrong design, EEGLAB plots and computes without an error, but for a question that is not yours.

The design also decides the kind of test: do the values come from the same participants (paired) or from different participants (unpaired)? This changes the statistical result a lot.

How do I do it in EEGLAB?

About the data (D6): the same animal_mine.study from the creation lesson. Every participant saw both kinds of picture (animal and distractor), so picture type is within subjects. Each participant has one gender (7 women and 7 men), so gender is between subjects. Together: a 2×2 mixed design.

1. Look at the design EEGLAB created

1. Run the “Starting point” script if the STUDY is not open.
2. Open Study › then Select/Edit study design(s). The Edit STUDY design — pop_studydesign() window opens.
3. The Design name list holds one design, STUDY.design 1. EEGLAB created it by itself when the STUDY was created.
4. The Edit the independent variables for this design part has two lines: Categorical variable: condition - Values (animal - distractor) and Categorical variable: group - Values (F - M). EEGLAB puts every standard variable with more than one value into the first design.

2. Create design 2 with a clear name

1. Press New next to the design list. Design 2 appears with no variables.
2. Press Rename, type image x gender in Study design name:, then Ok.
3. In the variables part press New. The Add variable window opens.
4. In Select independent variable choose condition. Both values, animal and distractor, are selected in Select variable values. Keep them selected, then Ok.
5. Press New again, choose group, keep F and M selected, then Ok.

3. Create design 3 with one variable

1. Press New next to the design list, then Rename it image only.
2. Add one variable only: condition with both its values.

4. Choose the current design and save

1. Click image x gender in the Design name list. The design selected when you press Ok becomes the current design.
2. Keep Re-save STUDY file (if saved previously) ticked, so animal_mine.study is saved with its three designs.
3. Press Ok. The command window shows Selecting STUDY design 2, then Resaving STUDY.

How do I write it as code?

% STUDY and ALLEEG come from the starting point script
% Design 2: image (within subjects) x gender (between subjects)
STUDY = std_makedesign(STUDY, ALLEEG, 2, 'name', 'image x gender', ...
    'variable1', 'condition', 'values1', {'animal', 'distractor'}, 'vartype1', 'categorical', ...
    'variable2', 'group', 'values2', {'F', 'M'}, 'vartype2', 'categorical');

% Design 3: image only, all participants in one group
STUDY = std_makedesign(STUDY, ALLEEG, 3, 'name', 'image only', ...
    'variable1', 'condition', 'values1', {'animal', 'distractor'}, 'vartype1', 'categorical', ...
    'variable2', 'none');

% Make design 2 the current design and save
STUDY = std_selectdesign(STUDY, ALLEEG, 2);
STUDY = pop_savestudy(STUDY, ALLEEG, 'savemode', 'resave');

% Every design with its variables, values and pairing
for k = 1:numel(STUDY.design)
    d = STUDY.design(k);
    fprintf('Design %d: %s\n', k, d.name);
    for v = 1:numel(d.variable)
        if ~isempty(d.variable(v).label)
            fprintf('   %-10s paired: %-4s values: %s\n', d.variable(v).label, ...
                d.variable(v).pairing, strjoin(d.variable(v).value, ' '));
        end
    end
end
fprintf('Current design: %d\n', STUDY.currentdesign);
eeglab redraw;

Which value should I choose?

Within subjects or between subjects?

EEGLAB decides paired or unpaired by itself from the data: if every participant has every value, it is paired. The script prints the decision in the line paired: on or paired: off.

Which design for which question?

How do I check it worked?

• The Design name list holds three designs: STUDY.design 1, image x gender and image only.
• The script prints for design 2 the two lines condition paired: on and group paired: off.
• The script prints Current design: 2.
• After File › then Clear study / Clear all and opening animal_mine.study again, the three designs are still in the window.

What mistakes do people make?

• Forgetting to choose the current design. Plots and statistics work on the design selected in the list at the last Ok. The script prints Current design.
• Putting a between-subject variable in condition. If you type gender as a condition, EEGLAB treats it as within subjects. The paired line reveals this: gender must be off.
• One design for every question. The 2×2 design answers the interaction, but the picture effect alone is simpler to read in a one-variable design: two curves instead of four, and a direct paired test.
• Selecting too few values. If you select only animal in Select variable values, there is nothing to compare. Make sure both values are selected.
• Thinking a new design needs a new precomputation. Since EEGLAB 2019 trials are saved one by one, and the design selects them at plot time. Do not precompute again because of a new design.

How do I write it in my paper?

A template: edit the data, settings and versions to match what you ran.

A 2 × 2 mixed design was defined in the STUDY, with image category (animal, distractor) as a within-subject factor and gender ([7] female, [7] male) as a between-subject factor (std_makedesign, EEGLAB 2026.1).

Going further

• EEGLAB Tutorial: https://eeglab.org/tutorials/10_Group_analysis/working_with_study_designs.html (10. Group analysis > c. STUDY designs)
• EEGLAB Tutorial: https://eeglab.org/tutorials/11_Scripting/command_line_study_functions.html (11. Scripting > Creating a STUDY)
• Introduction to the EEGLAB STUDY and Study Design by Arnaud Delorme: https://www.youtube.com/watch?v=HAGrNeh-kgA (Full video)
• Go-nogo categorization and detection task (OpenNeuro ds002680): https://openneuro.org/datasets/ds002680 (participants.tsv)