HHP 300: Module 1 - Lecture 1

What is data?

Observations collected from notes, surveys, and experiments that form the backbone of a statistical investigation

What is information?

Processed data

What is evidence?

Applied data

What is statistics?

Study of how to best collect, analyze, and draw conclusions from data

Parts of a statisitical test - 8

• Objective: what are you evaluating?

• Participant pool: overall people recruited

• Actual participants: who is part of the study/get data from?

  • Diagnostic criteria

• Treatment: the experiment

• Control

• Results

• Conclusion

  • Does the data show a “real” difference between the groups?

    • Observed difference between groups due to natural variation or real

• Can it be generalized to the population?

What is a data matrix? Where are the variables and observations located?

• Tidy’s data

• Variable: columns -

• Observations: rows |

What is a variable?

A characteristics measured by researchers

What are the two types of variables?

1. Categorical

2. Numerical

What is a categorical variable and its two sub-types?

Values that are names or labels (ex. color of a shoe)

• Ordinal: some natural ordering of levels (ex. education)

• Nominal: no natural ordering of levels (ex. gender)

What is a numerical variable and its two sub-types?

Numerical values where numerical operations are possible (ex. sum, differences, -/-, x)

• Discrete: only integers, countable (ex. # of family members)

What are cases?

Subjects of a study whom these variables are being measured

What are quantitative variables?

Numerical variables that represent measurable quantities

• Used in mathematical operations

All quantitative variables are ___, but not all numerical variables are ____. Give 2 examples

Numerical, quantitative

Ex.

• Coded categorical (ex. assigning 1 for low and 2 for high)

• A variable using #’s but it doesn’t rep a measurable quantity (ex. zip code, jersey #)

What is a measurement scale?

Qualify or quantify data variables, determines the kind of techniques to be used for analysis

• Type of data being collected determines the type of scale to be used

What are the four types of measurement scales?

Nominal, ordinal, interval, and ratio

What is a nominal scale? Give examples

Mutually exclusive categories within a variable; cannot be ordered or measured in a meaningful way

Ex. mode of transportation, color, gender, political party

What is a binary or dichotomous variable?

A categorical variable that has 2 possible values (yes/no)

What is an ordinal scale? Give examples

Categories within a variable that has a natural rank order

• Ex. language levels, socio-economic status, agreement lvl

How are ordinal variables usually assessed? Are averages always taken?

• Assessed using closed-ended survey questions that give participants several possible answers to choose from

• Averages aren’t always taken (ex. how important do you think is __)

**What is the interval scale? Give examples

Has a metric function, ordering, and meaningful operations among units, but with NO natural origin/true zero

• Data pts on the scale have the same difference between them (ex. difference between 10 and 20 & 50 and 60 is the same)

• True example: temp

What is the ratio scale? Give examples

Same as interval, except it has a true zero; variables can be meaningfully added, subtracted, multiplied, and divided

Ex. weight, height, length, ages in years

**How does the ordinal scale differ from the interval and ratio scale?

Doesn’t have category widths that represent equal increments of the underlying attribute

**What is ratio data like?

Continuous, with the potential of taking on infinite values and measurements can be made to many decimal places

Ex. weight in grams, BP, length in centimeters, and age in yrs

What is a scatterplot?

Relationship between 2 numerical variables

What is a associated/dependent variable?

Shows some connection with each other

What is an independent variable?

No connection between the 2

**Does association imply causation? How can causation be inferred?

Association doesn’t imply causation. Causation can be inferred from a randomized experiment

**Association doesn’t equal causation

1. Has the study been reproduced

2. Study timing/reverse causation: if A causes B, then A must come before B

3. No dose-response relationship: more exposure = greater effect

**What does it mean if there’s no mechanism explained?

Even if a study finds strong statistical association, causation is far more believable when there’s a know biological mechanism