# Expected Outputs Reference — Superpowered Instructor's Kit Date: 2026-11-15. 38 tasks. Values for input datasets are not invented; exact printed results shown only for self-contained deterministic exercises. Plot tasks describe the expected visual. R execution has not been verified in this authoring environment. ## W1-M01 — R as calculator **Kind:** Exact, deterministic example ```text [1] 7 [1] 8 [1] 11 ``` **Check:** Do not interpret R’s console prefix [1] as the result. ## W1-M02 — Assignment and reusable results **Kind:** Exact values where deterministic; otherwise preview ```text [1] 5.5 # ls() also lists current object names; that list depends on the R session. ``` **Check:** ls() is session-dependent and may include other objects. ## W1-M03 — Text, numbers, vectors and indexing **Kind:** Exact, deterministic example ```text [1] 6 [1] 8 7 [1] 6.5 ``` **Check:** In sleep_hours[sleep_hours > 6], the condition is applied element by element. ## W1-M04 — Built-in USArrests data frame **Kind:** Exact values where deterministic; otherwise preview ```text [1] 50 4 [1] "Murder" "Assault" "UrbanPop" "Rape" # head(crime) prints six state rows with four columns. ``` **Check:** Built-in datasets are useful for syntax practice, not evidence about present-day conditions. ## W1-M05 — Functions, help and extraction **Kind:** Exact values where deterministic; otherwise preview ```text [1] 6 # print(crime$Murder) shows all 50 values (truncated here). [1] 17.4 [1] 337 ``` **Check:** A maximum is a descriptive statistic; it does not identify its row unless you inspect which state attained it. ## W1-M06 — Removing objects **Kind:** Exact, deterministic example ```text [1] FALSE ``` **Check:** Running this section twice without recreating crime can fail because rm(crime) cannot find it. ## W1-E01 — Additional: reproducible report **Kind:** Exact, deterministic example ```text trial rt_ms 1 1 620 2 2 610 3 3 590 4 4 605 ``` **Check:** Four simulated trials do not support general population conclusions. ## W2-M01 — Packages and importing CSV **Kind:** Output format / checks (not fabricated values) ```text n_participants n_responses # Numeric counts depend on the included fictional public CSV files. ``` **Check:** Extract the full package and keep the data/ directory adjacent to the script. ## W2-M02 — Join by a compound key **Kind:** Output format / checks (not fabricated values) ```text rows unique_ids # Two named counts; values depend on the loaded public datasets. ``` **Check:** The nrow inequality catches row explosions but does not prove that all records matched correctly. ## W2-M03 — select(): columns **Kind:** Exact, deterministic example ```text [1] "id" "intervention" "ahiTotal" "cesdTotal" "sex" [6] "age" "educ" "income" "occasion" ``` **Check:** select() preserves row count and existing order unless other operations are added. ## W2-M04 — filter(): observations and a source correction **Kind:** Output format / checks (not fabricated values) ```text 0 1 # Counts for baseline and posttest; public fictional data values vary. ``` **Check:** Do not use 1/2 as pre/post: the original worksheet shortcut contradicts the codebook. ## W2-M05 — arrange(): rows **Kind:** Output format / checks (not fabricated values) ```text id occasion ahiTotal # First six rows after ascending sorting; exact values are data-dependent. ``` **Check:** Sorting and filtering are different: arrange() retains all input rows. ## W2-M06 — summarise(): overall means **Kind:** Output format / checks (not fabricated values) ```text n mean_ahi mean_cesd # One row with three summary columns; numbers depend on input CSVs. ``` **Check:** n() counts all rows, including those with an NA in a measured variable. ## W2-M07 — group_by(): distinct occasion summaries **Kind:** Output format / checks (not fabricated values) ```text occasion n mean_ahi mean_cesd # One summary row per occasion (0 and 1). ``` **Check:** If a score contains missing values, mean() without na.rm=TRUE may return NA. ## W2-M08 — mutate(): new column and scale caution **Kind:** Output format / checks (not fabricated values) ```text sadness cesdTotal ahiTotal # First six rows; sadness = cesdTotal - ahiTotal. ``` **Check:** This 'sadness' variable is ONLY a syntax exercise; it is not a legitimate clinical index. ## W2-M09 — Recode sex using the codebook **Kind:** Output format / checks (not fabricated values) ```text Female Male # Category frequencies; an NA row appears if unmatched/missing values occur. ``` **Check:** Reversing the codebook mappings would silently mislabel participants. ## W2-E01 — Additional: paired change and attrition **Kind:** Output format / checks (not fabricated values) ```text paired_n mean_change # One row: count of matched pairs and their average ahi_post - ahi_pre. ``` **Check:** Check uniqueness of (id, intervention, occasion); missing follow-ups and NA scores need explicit handling. ## W3-M01 — Read AQ, formats, scoring **Kind:** Output format / checks (not fabricated values) ```text participants questions rules # Three named counts; original private AQ source had 66 respondents and 10 items. ``` **Check:** The public data are fictional. Do not assume the original Moodle row counts are reproduced there. ## W3-M02 — Wide-to-long pivot **Kind:** Output format / checks (not fabricated values) ```text [1] # In original private data, there were 66 × 10 = 660 long-form rows. ``` **Check:** all_of() requires all Q1–Q10 columns to exist; verify column names and each participant's expected ten items. ## W3-M03 — Join directions and exact scoring rules **Kind:** Output format / checks (not fabricated values) ```text Id Question Response QFormat Score # First six matched response rows, each with a binary Score (0 or 1). ``` **Check:** inner_join can drop unmatched answers; equal row counts alone do not prove one-to-one matching. ## W3-M04 — Participant total AQ score **Kind:** Output format / checks (not fabricated values) ```text Id AQTotal # First six participants with integer totals from 0 to 10. ``` **Check:** A total within 0–10 may still come from fewer than ten scored questions. Check item counts per Id. ## W3-M05 — Pipe version of a complete workflow **Kind:** Exact, deterministic example ```text # No printed output when stopifnot(identical(...)) passes. # If it fails, R stops with an error. ``` **Check:** identical() checks classes, row order and attributes as well as values; equivalent data can sometimes differ in attributes. ## W3-M06 — Descriptives, cut-off and counts **Kind:** Output format / checks (not fabricated values) ```text n mean sd min max above6 # One summary row; followed by rows with AQTotal > 6. ``` **Check:** A questionnaire screening threshold must not be interpreted as a clinical diagnostic finding. ## W3-M07 — AQ frequency histogram **Kind:** Expected visual output ```text [Plot] Ten-point integer-scale histogram; x-axis 0–10, y-axis participant counts. ``` **Check:** The familiar normal '68% within 1 SD' rule need not hold for a bounded discrete score. ## W3-E01 — Additional: variance, SE, missingness **Kind:** Output format / checks (not fabricated values) ```text sample_variance standard_error # Two positive numeric values; data-dependent. ``` **Check:** Variance, SD and SE answer different questions. Do not substitute SE for SD on participant-level charts. ## W4-M01 — Histogram and binwidth **Kind:** Expected visual output ```text [Plot 1] AHI histogram with binwidth 5. [Plot 2] The same distribution with binwidth 1. ``` **Check:** Histogram appearance depends strongly on breaks; avoid claiming a shape is inherent to the raw data. ## W4-M02 — Scatterplot and sex legend **Kind:** Expected visual output ```text [Plot] AHI on x-axis; CES-D on y-axis; coloured points and a Female/Male legend. ``` **Check:** Repeated observations for the same person can make apparent independence misleading. ## W4-M03 — Boxplot, violin, and raw points **Kind:** Expected visual output ```text [Plot 1] AHI boxplot by occasion. [Plot 2] Violin + narrow boxplot + jittered individual points. ``` **Check:** A standard boxplot does not show mean ± SD; and a violin density may exaggerate thin groups. ## W4-M04 — Mean bar plot, grouped by intervention **Kind:** Expected visual output ```text [Plot 1] One mean-AHI bar per occasion. [Plot 2] Side-by-side intervention mean bars per occasion. ``` **Check:** Compare denominators n by group, and do not mistake a descriptive bar difference for a tested effect. ## W4-M05 — Line plot and SD error bars **Kind:** Expected visual output ```text [Plot] Intervention mean-AHI trajectories across occasions with vertical ±1 SD error bars. ``` **Check:** coord_cartesian() zooms without filtering original data; SD bars must not be called confidence intervals. ## W4-M06 — Themes, labels, facets and point style **Kind:** Expected visual output ```text [Plot 1] Classic scatter with title. [Plot 2] Black-and-white faceted scatter by sex. [Plot 3] Purple triangle scatter. ``` **Check:** Visual style is not evidence of a statistical effect; transparency influences visibility. ## W4-E01 — Additional: correlation and ATSQ context **Kind:** Output format / checks (not fabricated values) ```text correlation # One coefficient in [-1,1]. atsq_rows background_rows joined # Three data-dependent row counts. ``` **Check:** Correlation does not prove causality; missing pairs and repeated observations require care. ## W5-E01 — Sampling distributions and descriptive variation **Kind:** Exact, deterministic example ```text n = 12 mean_ms = 606.75 sd_ms ≈ 10.3320 se_ms ≈ 2.9826 ``` **Check:** Fictional trial values illustrate estimation only, not generalisable experimental results. ## W5-E02 — Resampling and bootstrap percentile interval **Kind:** Output format / checks (not fabricated values) ```text sd_n4 sd_n10 # Two simulation SDs (typically sd_n10 < sd_n4). 2.5% 97.5% # Two bootstrap percentile endpoints; exact values require running R. ``` **Check:** The seed makes R reproducible, but no numeric bootstrap result is claimed without executing the R source. ## W5-E03 — Visualising the resampling distributions **Kind:** Expected visual output ```text [Plot] Two side-by-side resampled-mean histograms: n=4 and n=10. ``` **Check:** The two histograms need comparable axes if you want a strong visual claim about width. ## W6-E01 — Null hypothesis and one-sample t test **Kind:** Exact values where deterministic; otherwise preview ```text t ≈ 2.2631 df = 11 p ≈ 0.04484 # t.test() additionally prints the 95% confidence interval and sample mean. ``` **Check:** A small p-value does not establish substantive importance or guarantee the model's assumptions. ## W6-E02 — Confidence interval and effect size **Kind:** Exact values where deterministic; otherwise preview ```text ci_lower ≈ 600.1854 ci_upper ≈ 613.3146 cohens_d ≈ 0.6533 # stopifnot() prints nothing if the two CI calculations agree. ``` **Check:** An interval describes uncertainty about a mean under assumptions, not a probability that individual scores lie inside it. ## W6-E03 — Regression, correlation, model diagnostics **Kind:** Output format / checks (not fabricated values) ```text Estimate Std. Error t value Pr(>|t|) (Intercept) ... ... ... ... ahiTotal ... ... ... ... r = ... ; r_squared = ... # Coefficient values depend on the public dataset. ``` **Check:** Do not treat this basic repeated-record model as a causal or fully valid independent-observation analysis.