Appendix A — Data Dictionaries

Appendix A — Data Dictionaries#

This appendix provides the complete data dictionaries for the two primary datasets used throughout this textbook.

1. The Framingham Heart Study (Observational)#

Source: Derived from the Framingham Heart Study teaching dataset provided by the National Heart, Lung, and Blood Institute (NHLBI/NIH) for educational use. Study: Framingham Heart Study (Contract N01-HC-25195), Framingham, Massachusetts, USA. Teaching subset: n = 500 participants, baseline examination (Period 1), complete cases. File: framingham_teaching.csv (in data/ folder of the book repository)

Important: This is a teaching dataset. Specific methods were employed to ensure participant anonymity. It is not appropriate for research publication or clinical decision-making.

Load in R:

fram_data <- read.csv("data/framingham_teaching.csv")

Variable Dictionary (Framingham)#

Variable	Type	Level	Values / Units	Description
`RANDID`	Integer	Nominal	1001–1500	Participant ID — a label, not a quantity
`SEX`	Integer → Factor	Nominal	1=Male, 2=Female	Sex at examination
`EDUC`	Integer → Ordered Factor	Ordinal	1=0-11yrs, 2=HS/GED, 3=Some college, 4=College grad	Education level — ranked, unequal gaps
`AGE`	Integer	Ratio	32–70 years	Age at baseline examination
`CURSMOKE`	Integer → Factor	Nominal	0=Non-smoker, 1=Current smoker	Smoking status at examination
`CIGPDAY`	Integer	Ratio	0–60 cigarettes/day	Cigarettes smoked per day (0 for non-smokers)
`TOTCHOL`	Integer	Ratio	mg/dL	Serum total cholesterol
`SYSBP`	Decimal	Ratio	mmHg	Systolic blood pressure
`DIABP`	Decimal	Ratio	mmHg	Diastolic blood pressure
`BPMEDS`	Integer → Factor	Nominal	0=No, 1=Yes	Currently taking anti-hypertensive medication
`DIABETES`	Integer → Factor	Nominal	0=No, 1=Yes	Diabetes diagnosis at examination
`BMI`	Decimal	Ratio	kg/m²	Body mass index
`HEARTRTE`	Integer	Ratio	beats/min	Ventricular rate (heart rate)
`GLUCOSE`	Integer	Ratio	mg/dL	Casual serum glucose
`PREVCHD`	Integer → Factor	Nominal	0=No, 1=Yes	Prevalent coronary heart disease at baseline
`PREVHYP`	Integer → Factor	Nominal	0=No, 1=Yes	Prevalent hypertension at baseline
`ANYCHD`	Integer → Factor	Nominal	0=No event, 1=CHD event	Incident CHD during follow-up (any: MI, angina, CHD death)
`TIMECHD`	Integer	Ratio	Days	Days from baseline to first CHD event or end of follow-up
`DEATH`	Integer → Factor	Nominal	0=Survived, 1=Died	All-cause mortality during follow-up
`TIMEDTH`	Integer	Ratio	Days	Days from baseline to death or end of follow-up
`STROKE`	Integer → Factor	Nominal	0=No, 1=Stroke	Incident stroke during follow-up
`TIMESTRK`	Integer	Ratio	Days	Days from baseline to stroke or end of follow-up

Notes on Framingham Variable Use#

Survival variables (TIMEDTH + DEATH, TIMECHD + ANYCHD): TIMEDTH and TIMECHD are censored survival data. Participants who did not have the event by end of follow-up have their time recorded but the true event time is unknown (censored). These variables must be analysed using Kaplan-Meier + log-rank test (Chapter 9) — not with standard regression alone.

EDUC (ordinal): The four education levels are ranked but have unequal gaps. Always convert to an ordered factor in R (ordered=TRUE). Use median or mode for summaries, not mean.

Binary variables (CURSMOKE, DIABETES, ANYCHD, etc.): Stored as 0/1 integers but are nominal. Convert to factors in R before analysis. The mean of a 0/1 variable gives a proportion (e.g., mean ANYCHD = 0.31 means 31% had a CHD event).

R Code: Complete Framingham Data Preparation#

# Load
fram_data <- read.csv("data/framingham_teaching.csv")

# Convert all categorical variables to labelled factors
fram_data$SEX      <- factor(fram_data$SEX,      levels=c(1,2),   labels=c("Male","Female"))
fram_data$EDUC     <- factor(fram_data$EDUC,     levels=1:4, ordered=TRUE, labels=c("0-11yrs","HS_GED","Some_college","College_grad"))
fram_data$CURSMOKE <- factor(fram_data$CURSMOKE, levels=c(0,1),   labels=c("Non-smoker","Smoker"))
fram_data$BPMEDS   <- factor(fram_data$BPMEDS,   levels=c(0,1),   labels=c("No","Yes"))
fram_data$DIABETES <- factor(fram_data$DIABETES, levels=c(0,1),   labels=c("No","Yes"))
fram_data$PREVCHD  <- factor(fram_data$PREVCHD,  levels=c(0,1),   labels=c("No","Yes"))
fram_data$PREVHYP  <- factor(fram_data$PREVHYP,  levels=c(0,1),   labels=c("No","Yes"))
fram_data$ANYCHD   <- factor(fram_data$ANYCHD,   levels=c(0,1),   labels=c("No_CHD","CHD_event"))
fram_data$DEATH    <- factor(fram_data$DEATH,    levels=c(0,1),   labels=c("Survived","Died"))
fram_data$STROKE   <- factor(fram_data$STROKE,   levels=c(0,1),   labels=c("No","Yes"))

2. The Anorexia Clinical Trial (Experimental)#

Source: The MASS package in R, originally published in A Handbook of Small Data Sets (Hand et al., 1994). Study: A clinical trial investigating the efficacy of different psychological treatments for young female patients with anorexia nervosa. Subset: n = 72 participants. File: Built directly into R; no external download required.

Load in R:

library(MASS)
data(anorexia)

Variable Dictionary (Anorexia)#

Variable	Type	Level	Values / Units	Description
`Treat`	Factor	Nominal	Cont, CBT, FT	Treatment group (Control, Cognitive Behavioral Therapy, Family Therapy)
`Prewt`	Decimal	Ratio	lbs (pounds)	Patient’s body weight before the study period
`Postwt`	Decimal	Ratio	lbs (pounds)	Patient’s body weight after the study period

Notes on Anorexia Variable Use#

Paired Data (Prewt and Postwt): Because these represent the exact same patient measured at two different times, they must be analysed using paired statistical tests (e.g., Paired T-Test) when comparing baseline to follow-up (Chapter 6).

Creating a Weight Gain Variable: To analyze the effectiveness of the treatments across groups, you will often need to compute a new variable: Weight_Change = Postwt - Prewt.

Dataset Use by Chapter#

Chapter	Framingham Analysis	Anorexia Analysis
1	Variable classification; factor conversion	Observational vs. Experimental study design
2	Descriptive statistics (`SYSBP`, `AGE`, `BMI`)	—
3	SE and 95% CI (`TOTCHOL`)	—
4	Normality; distributions; incidence	—
5	One-sample t-test vs reference	—
6	Independent t-test (`SYSBP` by `SEX`)	Paired t-test (`Prewt` vs `Postwt`)
7	Chi-Square (`CURSMOKE` × `ANYCHD`)	One-Way ANOVA (`Weight_Change` by `Treat`)
8	Linear regression (`AGE` → `SYSBP`)	Multiple regression predicting `Postwt`
9	KM Survival Analysis (`TIMEDTH` by `CURSMOKE`)	Test Selection Cheat Sheets