Homework Checkers

Interactive webexercises answer checkers for every analysis type

Overview

The psych350lab package provides interactive homework checkers that students can use in Quarto HTML documents. Students enter their answers into fill-in-the-blank boxes or select from dropdown menus, and the boxes turn green for correct or red for incorrect.

These checkers rely on the webexercises and tinytable packages.

Setup

This vignette uses the homework-checker theme and webexercises support bundled inside the nebraska Quarto extension (_extensions/nebraska/hwchecker/). To use it in your own documents, make sure the _extensions/nebraska folder is in your project and add the hwchecker assets to your YAML header:

format:
  html:
    mainfont: Sora
    filters:
      - _extensions/nebraska/hwchecker/webexercises.lua
    theme:
      - default
      - _extensions/nebraska/hwchecker/_variables.scss
      - _extensions/nebraska/hwchecker/_mixins.scss
      - _extensions/nebraska/hwchecker/custom.scss
    css:
      - _extensions/nebraska/hwchecker/webex.css
    include-after-body:
      - _extensions/nebraska/hwchecker/webex.js

Every checker chunk must include results: asis so that the HTML renders correctly:

#| results: asis

General Workflow

All checkers follow the same two-step pattern:

1. Compute results using an *_answers() function
2. Create checker using a create_*_checker() function

# Step 1: Compute
result <- some_answers(data, ...)

# Step 2: Create checker
create_some_checker(result, ...)

How Rounding and Computation Work

The *_answers() functions return unrounded values from the underlying statistical tests. No rounding is applied at the computation step so that full precision is preserved. The create_*_checker() functions then round each statistic to the appropriate number of decimal places (e.g., 2 for means and standard deviations, 3 for correlations and p-values) before building the fill-in-the-blank widgets. This means the checker’s expected answers always match what students would report following APA formatting conventions.

All analyses follow SPSS defaults. Missing values coded as −99 (or other SPSS user-missing codes) are converted to NA when the data are read with haven::read_sav() and haven::zap_missing(), so they are automatically excluded from computations.

Shared Setup

All examples below use the Superman media dataset. Load the data once at the top of your document:

library(psych350lab)
library(dplyr)
data(superman, package = "psych350data")

# Derive decade from year (not in the base dataset)
superman <- superman |>
  mutate(
    decade = case_when(
      year < 1970 ~ 1,
      year < 2000 ~ 2,
      year < 2010 ~ 3,
      TRUE        ~ 4
    )
  )

1. Descriptive Statistics

Students enter Mean, SD, and SEM for each variable and identify whether the mean is interpretable.

Compute

# Select and filter your analysis variables
my_data <- superman |>
  select(num, year, clark_height_in, clark_grp, height_diff, height_gap) |>
  filter(!is.na(height_diff))

# Compute descriptive statistics
desc_result <- descriptives_answers(my_data,
  vars = c("num", "year", "clark_height_in", "clark_grp",
           "height_diff", "height_gap")
)

Checker

create_descriptives_checker(
  vars = c("num", "year", "clark_height_in", "clark_grp",
           "height_diff", "height_gap"),
  desc_results_list = desc_result,
  label          = "num",
  quantitative   = c("year", "clark_height_in", "height_diff"),
  binary         = "clark_grp",
  multi_category = "height_gap",
  var_labels = c(
    num            = "Film Number",
    year           = "Release Year",
    clark_height_in = "Clark Height (in)",
    clark_grp      = "Height Group",
    height_diff    = "Height Difference",
    height_gap     = "Height Gap Category"
  )
)

Variable	Mean	SD	SEM	Interpretable?
Film Number				No - this is a label, not a variable Yes - this is a quantitative variable Yes - this is a binary variable No - this is a multiple-category variable
Release Year				No - this is a label, not a variable Yes - this is a quantitative variable Yes - this is a binary variable No - this is a multiple-category variable
Clark Height (in)				No - this is a label, not a variable Yes - this is a quantitative variable Yes - this is a binary variable No - this is a multiple-category variable
Height Group				No - this is a label, not a variable Yes - this is a quantitative variable Yes - this is a binary variable No - this is a multiple-category variable
Height Difference				No - this is a label, not a variable Yes - this is a quantitative variable Yes - this is a binary variable No - this is a multiple-category variable
Height Gap Category				No - this is a label, not a variable Yes - this is a quantitative variable Yes - this is a binary variable No - this is a multiple-category variable

Variable Type Categories

Category	Meaning	Mean Interpretable?
label	ID variable (e.g., film number)	No
quantitative	Continuous numeric variable	Yes
binary	Dichotomous variable (2 levels)	Yes
multi_category	Nominal variable (3+ levels)	No

2. Correlations

Students report the Pearson r, p-value, degrees of freedom, and descriptive statistics for a bivariate correlation.

Compute

corr_result <- corr_answers(superman, "clark_height_in", "rt_critics_score")

Checker

create_corr_checker(
  rh_name = "RH1",
  vars = c("Clark Height", "Critics Score"),
  corr_results_list = corr_result
)

	r	p	df	Reject or Retain?
Correlation: RH1				Retain the H0 null hypothesis Reject the H0 null hypothesis
	Mean	SD	N
Variable 1: Clark Height
	Mean	SD	N
Variable 2: Critics Score

3. Chi-Square (2×2)

Students report observed frequencies, the chi-square statistic, p-value, and degrees of freedom for a 2×2 test of independence.

Compute

chi_result <- chi_square_answers(superman, "clark_grp", "tomatometer")

Checker

create_chisq_checker(
  rh_name      = "RH1",
  chi_results_list = chi_result,
  var1_labels  = c("Under 6ft", "6ft+"),
  var2_labels  = c("Rotten", "Fresh")
)

	χ²	p	df	Reject or Retain H0?
Chi-Square: RH1				Retain the H0 null hypothesis Reject the H0 null hypothesis
	n		n
Number of Under 6ft in the sample		Number of Rotten in the sample
Number of 6ft+ in the sample		Number of Fresh in the sample

4. K-Group Chi-Square (k×2)

When the row variable has three or more levels, a significant omnibus chi-square is followed by pairwise 2×2 comparisons. Two checkers are needed: one for the omnibus test and one for the pairwise follow-ups.

Compute

kgroup_chi <- chi_square_kgroup_answers(
  data       = superman,
  var1       = "decade",
  var2       = "tomatometer",
  var1_labels = c("1950s", "1970-80s", "2000s", "2010s"),
  var2_labels = c("Rotten", "Fresh")
)

Omnibus Checker

create_chisq_omnibus_table(
  rh_name            = "RH1",
  chisq_results_list = kgroup_chi,
  var1_labels        = c("1950s", "1970-80s", "2000s", "2010s"),
  var2_labels        = c("Rotten", "Fresh")
)

	χ²	p	df	N		Do we need to perform pairwise comparisons?
Chi-Square: RH1						No – a nonsignificant Omnibus χ² test Yes – significant Omnibus χ² test
Number of 1950s in sample		Number of 1970-80s in sample		Number of 2000s in sample
Number of Rotten in sample		Number of Fresh in sample

Pairwise Checker

create_chisq_pairwise_checker(
  chisq_results_list = kgroup_chi,
  var1_labels        = c("1950s", "1970-80s", "2000s", "2010s"),
  var2_labels        = c("Rotten", "Fresh")
)

Chi-Square critical
	% comparison	χ² Result	Type of Error	Effect Size (r)	Power Problem?
1950s vs 2010s	% vs %		Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
1970-80s vs 2010s	% vs %		Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
2000s vs 2010s	% vs %		Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)

5. One-Way Between-Groups ANOVA (2 groups)

Students report the F statistic, p-value, degrees of freedom, MSE, and group descriptives for a two-group between-groups comparison.

Compute

bg_result <- bg_anova_answers(superman, iv = "clark_grp", dv = "rt_critics_score")

Checker

create_bg_anova_checker(
  rh_name            = "RH1",
  vars               = c("clark_grp", "rt_critics_score"),
  anova_results_list = bg_result
)

	Type
ANOVA Type: RH1	Between-Groups (BG) Within-Groups (WG)
	F	p	df(between)	df(within)	MSE
BG ANOVA: RH1
	Reject or Retain H0?
Decision:	Retain the H0 null hypothesis Reject the H0 null hypothesis
	Mean	SD	n
6ft or taller
Under 6ft

6. One-Way Within-Groups ANOVA (2 conditions)

Students report the F statistic, p-value, degrees of freedom, MSE, and condition descriptives for a repeated-measures comparison.

Compute

wg_result <- wg_anova_answers(
  superman,
  dv1 = "rt_critics_score",
  dv2 = "rt_audience_score",
  dv1_label = "Critics Score",
  dv2_label = "Audience Score"
)

Checker

create_wg_anova_checker(
  rh_name            = "RH1",
  vars               = c("Critics Score", "Audience Score"),
  anova_results_list = wg_result,
  condition_labels   = c("Critics Score", "Audience Score")
)

	Type
ANOVA Type: RH1	Between-Groups (BG) Within-Groups (WG)
	F	p	df(effect)	df(error)	MSE
WG ANOVA: RH1
	Reject or Retain H0?
Decision:	Retain the H0 null hypothesis Reject the H0 null hypothesis
	Mean	SD	n
Critics Score
Audience Score

7. K-Group Between-Groups ANOVA (3+ groups)

When the IV has three or more groups, the omnibus F-test is followed by LSD pairwise comparisons. Two checkers are needed.

Compute

kgroup_bg <- anova_kgroup_answers(
  data         = superman,
  dv           = "rt_critics_score",
  iv           = "decade",
  group_labels = c("1950s", "1970-80s", "2000s", "2010s")
)

Omnibus Checker

create_anova_omnibus_checker(
  rh_name            = "RH1",
  anova_results_list = kgroup_bg,
  group_labels       = c("1950s", "1970-80s", "2000s", "2010s")
)

	F	p	df (between)	df (within)	MSE	Do we need to perform LSD pairwise comparisons?
BG ANOVA: RH1						No - a nonsignificant Omnibus F-test Yes - significant Omnibus F-test
	N	k	average n
	Mean	SD	n
1950s
1970-80s
2000s
2010s

LSD Pairwise Checker

create_lsd_pairwise_checker(
  anova_results_list = kgroup_bg,
  group_labels       = c("1950s", "1970-80s", "2000s", "2010s")
)

LSDmmd
	Mean Difference	LSD Result	Type of Error	Effect Size (r)	Power Problem?
1950s vs 1970-80s			Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
1950s vs 2000s			Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
1950s vs 2010s			Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
1970-80s vs 2000s			Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
1970-80s vs 2010s			Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)
2000s vs 2010s			Type I & III Type II		No - rejecting H0: means there was sufficient power No - effect is "too small to be interesting," (r < .10) Yes - The effect is "large enough to be interesting," (r > .10)

8. Factorial ANOVA (2×2 Between-Groups / Mixed)

Students report the interaction and main-effect F-tests, degrees of freedom, MSE, and follow-up comparisons. Two checkers cover the ANOVA results and the cell/marginal descriptive statistics.

Compute

sm_factorial <- superman |>
  mutate(era = if_else(year >= 2000, "Post-2000", "Pre-2000"))

factorial_result <- anova_factorial_answers(
  data      = sm_factorial,
  dv        = "rt_critics_score",
  iv1       = "clark_grp",
  iv2       = "era",
  iv1_labels = c("Under 6ft", "6ft+"),
  iv2_labels = c("Pre-2000", "Post-2000")
)

ANOVA Checker

create_factbg_anova_checker(
  rh_name            = "RH1",
  anova_results_list = factorial_result,
  iv1_name           = "Height Group",
  iv2_name           = "Era"
)

	F	p	df (between)	df (within)	MSE	Do we need to perform LSD pairwise comparisons?
Interaction: Height Group x Era						No - a nonsignificant interaction Yes - significant interaction
	# of conditions	average n	df error	MSe	LSDmmd
Components for LSDmmd:
	F	p	df (between)	df (within)	MSE
Main Effect: Height Group
	F	p	df (between)	df (within)	MSE
Main Effect: Era

Descriptives Checker (cell means and EMMs)

create_factbg_desc_checker(
  anova_results_list = factorial_result,
  iv1_name           = "Height Group",
  iv2_name           = "Era",
  iv1_labels         = c("Under 6ft", "6ft+"),
  iv2_labels         = c("Pre-2000", "Post-2000")
)

	Pre-2000	Post-2000	EMM
Under 6ft
6ft+
EMM

9. Linear Regression

Students report model-level statistics (R, R², F, df, p) and predictor-level statistics (r, b, significance, interpretation category). Two checkers split these.

Compute

sm_reg <- superman |>
  filter(
    !is.na(rt_critics_score),
    !is.na(clark_height_in),
    !is.na(rt_audience_score)
  )

reg_result <- linear_reg_answers(
  data              = sm_reg,
  criterion         = "rt_critics_score",
  quant_predictors  = c("clark_height_in", "rt_audience_score"),
  quant_labels      = c("Clark Height", "Audience Score"),
  criterion_label   = "Critics Score"
)

Model Summary Checker

create_regression_model_checker(reg_result)

	R	R²	F	df1, df2	p	Does the model work?
Model Summary				,		Yes No

Predictor Results Checker

create_regression_predictor_checker(reg_result, show_legend = TRUE)

Significance Key:

• ns = p > .05 (not significant)
• * = p < .05
• ** = p < .01
• *** = p < .001

Result Categories:

• a = Neither r nor b significant
• b = r & b both significant & same sign
• c = r significant but not b
• d = suppressor effect

Predictor	Type	r	r sig	b	b sig	Result
Clark Height	Binary Quant		ns * ** ***		ns * ** ***	a b c d
Audience Score	Binary Quant		ns * ** ***		ns * ** ***	a b c d

Quick Reference

Analysis Functions

Function	Analysis Type	Key Arguments
descriptives_answers()	Univariate stats	data, vars
corr_answers()	Correlation	data, var1, var2
chi_square_answers()	Chi-Square (2×2)	data, var1, var2
chi_square_kgroup_answers()	Chi-Square (k×2)	data, var1, var2, var1_labels, var2_labels
bg_anova_answers()	BG ANOVA (2 groups)	data, iv, dv
wg_anova_answers()	WG ANOVA (2 cond.)	data, dv1, dv2
anova_kgroup_answers()	BG ANOVA (k groups)	data, dv, iv, group_labels
anova_factorial_answers()	2×2 Factorial (BG)	data, dv, iv1, iv2
anova_factmg_answers()	2×2 Factorial (MG)	data, dv, iv1, iv2
anova_factwg_answers()	2×2 Factorial (WG)	data, dv_a1b1, dv_a1b2, dv_a2b1, dv_a2b2
linear_reg_answers()	Regression	data, criterion, quant_predictors

Checker Functions

Function	Analysis Type	Key Components
create_descriptives_checker()	Univariate	Mean, SD, SEM, interpretability
create_corr_checker()	Correlation	r, p, df, descriptives
create_chisq_checker()	Chi-Square (2×2)	χ², p, df, frequencies
create_chisq_omnibus_table()	Chi-Square omnibus	χ², p, df, N, frequencies
create_chisq_pairwise_checker()	Chi-Square pairwise	χ² critical, percentages, effect sizes
create_bg_anova_checker()	BG ANOVA (2 groups)	F, p, df, MSE, group stats
create_wg_anova_checker()	WG ANOVA (2 cond.)	F, p, df, MSE, condition stats
create_anova_omnibus_checker()	BG ANOVA (k groups)	F, p, df, MSE, N, k, group stats
create_lsd_pairwise_checker()	LSD pairwise	LSD MMD, mean differences, effect sizes
create_factbg_anova_checker()	2×2 Factorial (BG) ANOVA	Interaction, main effects, LSD
create_factbg_desc_checker()	2×2 Factorial (BG) desc.	Cell means, EMMs
create_factmg_anova_checker()	2×2 Factorial (MG) ANOVA	Interaction, main effects, LSD
create_factmg_desc_checker()	2×2 Factorial (MG) desc.	Cell means, EMMs
create_factwg_anova_checker()	2×2 Factorial (WG) ANOVA	Interaction, main effects, LSD
create_factwg_desc_checker()	2×2 Factorial (WG) desc.	Cell means, EMMs
create_regression_model_checker()	Regression model	R, R², F, df, p
create_regression_predictor_checker()	Regression predictors	r, b, significance, categories

Troubleshooting

Tables not rendering. Make sure your chunk includes #| results: asis.

webexercises styles missing. Check that your YAML includes the CSS and JS files as shown in the Setup section above.

Checker shows wrong answers. Make sure you are passing the result from the *_answers() function, not raw data.

Package not found errors. Install required packages with install.packages(c("tinytable", "webexercises")).