def test_paired_t_test():
data_path = get_data_path('spiderLong_within.csv')
# Declare and annotate the variables of interest
variables = [{
'name': 'Group',
'data type': 'nominal',
'categories': ['Picture', 'Real Spider']
}, {
'name': 'Anxiety',
'data type': 'ratio'
}]
experimental_design = {
'study type': 'experiment',
'independent variables': 'Group',
'dependent variables': 'Anxiety',
'within subjects': 'Group'
}
assumptions = {'Type I (False Positive) Error Rate': 0.05}
tea.data(data_path, key="id")
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['Group', 'Anxiety'], ['Group:Real Spider > Picture'])
# print("\nfrom Field et al.")
# print("Expected outcome: Paired/Dependent t-test")
print('++++++++++++')
def test_pearson_corr():
data_path = get_data_path('statex77.csv')
# data_path2 = get_data_path('statex87.csv')
# Declare and annotate the variables of interest
variables = [{
'name': 'Illiteracy',
'data type': 'interval',
'categories': [0, 100]
}, {
'name': 'Life Exp',
'data type': 'ratio',
}]
experimental_design = {
'study type': 'observational study',
'contributor variables': ['Illiteracy', 'Life Exp'],
'outcome variables': ''
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
'normal distribution': ['Illiteracy']
}
tea.data(data_path)
# tea.data(data_path2)
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions, 'strict')
results = tea.hypothesize(['Illiteracy', 'Life Exp'],
['Illiteracy ~ Life Exp'])
# print("\nfrom Kabacoff")
# print("Expected outcome: Pearson")
print('++++++++++++')
def test_wilcoxon_signed_rank():
data_path = get_data_path('alcohol.csv')
# Declare and annotate the variables of interest
variables = [{
'name': 'drug',
'data type': 'nominal',
'categories': ['Alcohol']
}, {
'name': 'day',
'data type': 'nominal',
'categories': ['sundayBDI', 'wedsBDI']
}, {
'name': 'value',
'data type': 'ratio'
}]
experimental_design = {
'study type': 'experiment',
'independent variables': 'day',
'dependent variables': 'value',
'within subjects': 'day'
}
assumptions = {'Type I (False Positive) Error Rate': 0.05}
tea.data(data_path)
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['day', 'value'], ['day:sundayBDI != wedsBDI'])
# print("\nfrom Field et al.")
# print("Expected outcome: Wilcoxon signed rank test")
print('++++++++++++')
def test_f_test():
data_path = get_data_path('cholesterol.csv')
# Declare and annotate the variables of interest
variables = [{
'name': 'trt',
'data type': 'nominal',
'categories': ['1time', '2times', '4times', 'drugD', 'drugE']
}, {
'name': 'response',
'data type': 'ratio'
}]
experimental_design = {
'study type': 'experiment',
'independent variables': 'trt',
'dependent variables': 'response',
'between subjects': 'trt'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path)
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['trt', 'response'])
# print("\nFrom Field et al.")
# print("Expected outcome: Oneway ANOVA (F) test")
print('++++++++++++')
def test_kruskall_wallis():
data_path = get_data_path('soya.csv')
# Declare and annotate the variables of interest
variables = [
{
'name' : 'Sperm',
'data type' : 'interval'
},
{
'name' : 'Soya',
'data type' : 'ordinal',
'categories': ['No Soya', '1 Soya Meal', '4 Soya Meals', '7 Soya Meals']
}
]
experimental_design = {
'study type': 'experiment',
'independent variables': 'Soya',
'dependent variables': 'Sperm',
'between subjects': 'Soya'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path)
tea.define_variables(variables)
tea.define_study_design(experimental_design) # Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.assume(assumptions)
tea.hypothesize(['Soya', 'Sperm'])
print("\nFrom Field et al.")
print("Expected outcome: Kruskall Wallis")
def test_wilcoxon_signed_rank_1():
tea.data('./tests/data/real_stats_1.csv')
variables = [{
'name': 'Subject',
'data type': 'ratio'
}, {
'name': 'Source',
'data type': 'nominal',
'categories': ['Memory', 'Median']
}, {
'name': 'Score',
'data type': 'ratio'
}]
experimental_design = {
'study type': 'experiment',
'independent variables': 'Source',
'dependent variables': 'Score',
'within subjects': 'Source'
}
assumptions = {'Type I (False Positive) Error Rate': 0.05}
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['Source', 'Score'], ['Source:Memory != Median'])
def test_wilcoxon_signed_rank_2():
tea.data('./tests/data/real_stats_2.csv', key='Couple')
variables = [{
'name': 'Couple',
'data type': 'ratio'
}, {
'name': 'Person',
'data type': 'nominal',
'categories': ['Wife', 'Husband']
}, {
'name': 'Score',
'data type': 'ratio'
}]
experimental_design = {
'study type': 'observational study',
'contributor variables': 'Person',
'outcome variables': 'Score',
'within subjects': 'Person'
}
assumptions = {'Type I (False Positive) Error Rate': 0.05}
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['Person', 'Score'], ['Person:Wife != Husband'])
def test_indep_t_test():
data_path = get_data_path('UScrime.csv')
# Declare and annotate the variables of interest
variables = [{
'name': 'So',
'data type': 'nominal',
'categories': ['0', '1']
}, {
'name': 'Prob',
'data type': 'ratio',
'range': [0, 1]
}]
experimental_design = {
'study type': 'observational study',
'contributor variables': 'So',
'outcome variables': 'Prob',
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
'groups normally distributed': [['Prob', 'So']]
}
transformations = {'log': ['Prob']}
tea.data(data_path)
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['So', 'Prob'], ['So:1 > 0']) # Southern is greater
# print("\nfrom Kabacoff")
# print("Expected outcome: Student's t-test")
print('++++++++++++')
def test_chi_square_with_dataframe():
data_path = get_data_path('catsData.csv')
data_frame = pd.read_csv(data_path)
# Declare and annotate the variables of interest
variables = [{
'name': 'Training',
'data type': 'nominal',
'categories': ['Food as Reward', 'Affection as Reward']
}, {
'name': 'Dance',
'data type': 'nominal',
'categories': ['Yes', 'No']
}]
experimental_design = {
'study type': 'observational study',
'contributor variables': 'Training',
'outcome variables': 'Dance'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_frame) # Passes data_frame instead of data_path
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['Training', 'Dance'])
# print('Chi square')
print('++++++++++++')
def test_pearson_corr_2():
data_path = get_data_path('exam.csv')
# Declare and annotate the variables of interest
variables = [
{
'name' : 'Exam',
'data type' : 'ratio',
'range' : [0, 100]
},
{
'name' : 'Anxiety',
'data type' : 'interval',
'range' : [0, 100]
},
{
'name' : 'Gender',
'data type' : 'nominal',
'categories' : ['Male', 'Female']
},
{
'name' : 'Revise',
'data type' : 'ratio'
}
]
experimental_design = {
'study type': 'observational study',
'contributor variables': ['Anxiety', 'Gender', 'Revise'],
'outcome variables': 'Exam'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path)
tea.define_variables(variables)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
results = tea.hypothesize(['Anxiety', 'Exam'])
results = tea.hypothesize(['Revise', 'Exam'])
results = tea.hypothesize(['Anxiety', 'Revise'])
print("\nfrom Field et al.")
print("Expected outcome: Pearson")
def test_two_way_anova():
data_path = get_data_path('co2.csv')
# Declare and annotate the variables of interest
variables = [{
'name': 'uptake',
'data type': 'interval'
}, {
'name': 'Type',
'data type': 'nominal',
'categories': ['Quebec', 'Mississippi']
}, {
'name': 'conc',
'data type': 'ordinal',
'categories': [95, 175, 250, 350, 500, 675, 1000]
}]
experimental_design = {
'study type': 'experiment',
'independent variables': ['Type', 'conc'],
'dependent variables': 'uptake',
'within subjects': 'conc',
'between subjects': 'Type'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
'groups normally distributed': [['Type', 'uptake'], ['Type', 'conc']],
'equal variance': [['Type', 'uptake'], ['conc', 'uptake']]
}
tea.data(data_path)
tea.define_variables(variables)
tea.define_study_design(
experimental_design
) # Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.assume(assumptions, mode='relaxed')
tea.hypothesize(['uptake', 'conc',
'Type']) # Fails: not all groups are normal
#Type main effect?
# print('Supposed to be 2 way ANOVA')
print('++++++++++++')
def test_kendall_tau_corr():
data_path = get_data_path('liar.csv')
# Declare and annotate the variables of interest
variables = [
{
'name': 'Creativity',
'data type': 'interval'
},
{
'name': 'Position',
'data type': 'ordinal',
'categories': [6, 5, 4, 3, 2, 1] # ordered from lowest to highest
},
{
'name': 'Novice',
'data type': 'nominal',
'categories': [0, 1]
}
]
experimental_design = {
'study type': 'observational study',
'contributor variables': ['Novice', 'Creativity'],
'outcome variables': 'Position'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path)
tea.define_variables(variables)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
results = tea.hypothesize(
['Position', 'Creativity'],
['Position:1 > 6', 'Position:1 > 2']) # I think this works!?
# print("\nfrom Field et al.")
# print("Expected outcome: Kendall Tau")
print('++++++++++++')
def test_rm_one_way_anova():
data_path = get_data_path('co2.csv')
# Declare and annotate the variables of interest
variables = [
{
'name' : 'uptake',
'data type' : 'interval'
},
{
'name' : 'Type',
'data type' : 'nominal',
'categories': ['Quebec', 'Mississippi']
},
{
'name' : 'conc',
'data type' : 'ordinal',
'categories': [95, 175, 250, 350, 500, 675, 1000]
}
]
experimental_design = {
'study type': 'experiment',
'independent variables': ['Type', 'conc'],
'dependent variables': 'uptake',
'within subjects': 'conc',
'between subjects': 'Type'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path, key="Plant")
tea.define_variables(variables)
tea.define_study_design(experimental_design) # Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.assume(assumptions)
tea.hypothesize(['uptake', 'conc'])
print("\nFrom Field et al.")
print("Expected outcome: Repeated Measures One Way ANOVA")
def test_factorial_anova():
data_path = get_data_path('gogglesData.csv')
# Declare and annotate the variables of interest
variables = [
{
'name' : 'gender',
'data type' : 'nominal',
'categories' : ['Female', 'Male']
},
{
'name' : 'alcohol',
'data type' : 'nominal',
'categories': ['None', '2 Pints', '4 Pints']
},
{
'name' : 'attractiveness',
'data type' : 'interval'
}
]
experimental_design = {
'study type': 'experiment',
'independent variables': ['gender', 'alcohol'],
'dependent variables': 'attractiveness',
'between subjects': ['gender', 'alcohol']
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path)
tea.define_variables(variables)
tea.define_study_design(experimental_design) # Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.assume(assumptions)
tea.hypothesize(['attractiveness', 'gender', 'alcohol'])
# alcohol main effect?
print("\nFrom Field et al.")
print("Expected outcome: Factorial ANOVA")
def test_pointbiserial_corr():
data_path = get_data_path('pbcorr.csv')
# Declare and annotate the variables of interest
variables = [
{
'name': 'time',
'data type': 'ratio'
},
{
'name': 'gender',
'data type': 'nominal',
'categories': [0, 1] # ordered from lowest to highest
},
{
'name': 'recode',
'data type': 'nominal',
'categories': [0, 1]
}
]
experimental_design = {
'study type': 'observational study',
'contributor variables': ['gender', 'recode'],
'outcome variables': 'time'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path)
tea.define_variables(variables)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
# I think this works!?
tea.hypothesize(['time', 'gender'], ['gender:1 > 0'])
# print("\nfrom Field et al.")
# print("Expected outcome: Pointbiserial")
print('++++++++++++')
def test_mann_whitney_0():
tea.data('./tests/data/real_stats_3.csv')
variables = [{
'name': 'Treatment',
'data type': 'nominal',
'categories': ['Control', 'Drug']
}, {
'name': 'Score',
'data type': 'ratio'
}]
experimental_design = {
'study type': 'experiment',
'independent variables': 'Treatment',
'dependent variables': 'Score'
}
assumptions = {'Type I (False Positive) Error Rate': 0.05}
tea.define_variables(variables)
# Allows for using multiple study designs for the same dataset (could lead to phishing but also practical for saving analyses and reusing as many parts of analyses as possible)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
tea.hypothesize(['Treatment', 'Score'], ['Treatment:Control != Drug'])
import pandas as pd
import tea
## Preregistration: https://osf.io/bmer9/
## Split Dataset at https://github.com/jhofman/visual-effects-chi2020/tree/master
# df = pd.read_csv('data/df.csv', index_col=0)
# for i, g in df.groupby('text_condition'):
# name = 'data/' + i + '.csv'
# g.to_csv(name)
#################################################
## Run Tea on separate conditions
tea.data("data/show_viz_stats_only.csv", key="worker_id")
variables = [
{
'name': 'condition',
'data type': 'nominal',
'categories': ['SE', 'SD']
},
{
'name': 'text_condition',
'data type': 'nominal',
'categories': ['show_both_stats', 'show_viz_stats_only']
},
{
'name': 'wtp_final',
'data type': 'interval'
},
{
'name': 'superiority_special',
'name': 'BDIchange',
'data type': 'ratio'
}
]
study_design = {
'study type': 'observational study',
'contributor variables': ['drug', 'sundayBDI'],
'outcome variables': ['BDIchange', 'wedsBDI']
}
assumptions = {
'Type I (False Positive) Error Rate': 0.01
}
tea.data(df)
tea.define_variables(variables)
tea.define_study_design(study_design)
tea.assume(assumptions)
tea.hypothesize(['drug', 'wedsBDI'], ['drug:Ecstasy > Alcohol'])
tea.hypothesize(['sundayBDI', 'BDIchange'], ['sundayBDI ~ BDIchange'])
'''
Results:
--------------
Test: pointbiserial_corr_a
***Test assumptions:
Exactly two variables involved in analysis: drug, wedsBDI
Continuous (not categorical) data: wedsBDI
Normal distribution: wedsBDI: NormalTest(W=0.7817826867103577, p_value=0.04004703089594841)
Variable is categorical: drug
}, {
'name': 'uptake',
'data type': 'ratio'
}]
study_design = {
'study type': 'observational study',
'contributor variables': 'Plant',
'outcome variables': 'uptake'
}
assumptions = {
'Type I (False Positive) Error Rate': 0.05,
}
tea.data(data_path, key='Id')
tea.define_variables(variables)
tea.define_study_design(study_design)
tea.assume(assumptions)
tea.hypothesize(['Plant', 'uptake'], ['Plant: Qn1 < Qn2', 'Plant: Qc2 < Qc3'])
'''
Results:
--------------
Test: kruskall_wallis
***Test assumptions:
Independent (not paired) observations: Plant
Exactly one explanatory variable: Plant
Exactly one explained variable: uptake
Continuous (not categorical) data: uptake
Variable is categorical: Plant
Variable has two or more categories: Plant
import tea
tea.data("./UScrime.csv")
variables = [
{
'name': 'So',
'data type': 'nominal',
'categories': ['0', '1']
},
{
'name': 'Prob',
'data type': 'ratio',
'range': [0, 1] # optional
},
{
'name': 'Ineq',
'data type': 'ratio'
}
]
study_design = {
'study type': 'observational study',
'contributor variables': ['So', 'Prob'],
'outcome variables': ['Prob', 'Ineq']
}
assumptions = {
# 'equal variance': [['So', 'Ineq']],
# 'groups normally distributed': [['So', 'Prob']],
'Type I (False Positive) Error Rate': 0.05
'categories': ['AR', 'TV']
}, {
'name': 'Score',
'data type': 'ordinal',
'categories': [1, 2, 3, 4, 5]
}]
experimental_design = {
'study type': 'experiment',
'independent variables': 'Condition',
'dependent variables': 'Score'
}
assumptions = {'Type I (False Positive) Error Rate': 0.01969}
tea.data(data_path, key='ID')
tea.define_variables(variables)
tea.define_study_design(experimental_design)
tea.assume(assumptions)
results = tea.hypothesize(['Score', 'Condition'], ['Condition:AR > TV'])
'''
Results:
--------------
Test: mannwhitney_u
***Test assumptions:
Exactly one explanatory variable: Condition
Exactly one explained variable: Score
Independent (not paired) observations: Condition
Variable is categorical: Condition
Variable has two categories: Condition
Continuous OR ORDINAL (not nominal) data: Score
