Module Components

• Overview

Topics

• Datafiles and Programs to Run Statistics
• Variables, Values, and Codebooks
• Types of Variables & Levels of Measurement
• Missing Data
• Significant Figures and Rounding

NRSG 795: BIOSTATISTICS FOR EVIDENCE-BASED PRACTICE

Module 1: Variables, Values, and Spreadsheets as Databases

Types of Variables & Levels of Measurement

When we explore causal relationships we usually use terms to distinguish the ‘cause’ from the ‘outcome’.

• Independent variable - is the presumed cause or source of influence. In a clinical trial the receipt of an intervention or not would be an example of this kind of variable. Other names used are predictor or antecedent (use of these terms should be restricted to longitudinal types of designs as they imply causality), covariate (implies more than one independent variable), and it is often referred to as the ‘x’ variable in equations.
• Dependent variable - is the consequence or the presumed effect resulting from or having a relationship with the independent variable. In a clinical trial this would be the care variable that you are trying to influence. Other names used are outcome and it is often referred to as the ‘y’ variable in equations.

The numerical values are just short codes for the longer names and allow us to perform various analyses depending on the type of variable. A fundamental characteristic of variables that helps determine what statistical procedure to employ is the level of measurement. There are four levels of measurement. At lower levels of measurement, assumptions tend to be less restrictive and data analyses tend to be less sensitive. The level of measurement of the outcome (or dependent variable) is used to determine what statistical test to perform.

• Categorical or nominal variables are those that have no intrinsic ordering. For example, gender has two categories (male and female). While numbers are often assigned to these responses (e.g., 1=male, 0=female), there is no intrinsic ordering to the categories-the numbers are just arbitrary tags or values. A special type of nominal variable is when is it dichotomous-meaning it has only two values (yes/no, male/female, true/false).
• Ordinal variables are similar to categorical but they have some order that can be justified. For example, education can be ordered by high school, college, and graduate.

For nominal and ordinal variables, the most common descriptive statistics are frequencies and proportions.

• Interval or ratio level data have intervals between the variables that are equally spaced such that the units between are the same at each interval. Ratio level is an interval level with the additional property that its zero position indicates the absence of the quantity being measured. The interval and ratio level data have a distribution and that variation in the values (e.g., weight in pounds) allows analysis of the variation through more complex statistical procedures. These kinds of variables are also often referred to as continuous variables.

Three characteristics that are commonly described for interval or ratio level variables:

• central tendency (measured by mean, median, mode)
• variability (measured by variance, standard deviation, interquartile range)
• shape of distribution (measured as normal or skewed)

Variables are how we operationalize the concepts being study. Articles should have detailed descriptions of the variables studied so that you can determine the level of measurement.  For example, individual measures that span a range of values (e.g., 1-10 with information telling you what the scores mean [low-high]; a summated scale score) can be interpreted as an interval level measure. Your analysis assignment scores are an example of an interval variable that will have values between 0-10. You can score any amount of points between 0 and 10; some will get a score of 9.5, 8.9 etc.  A pain score when asked to rate one’s pain level between 1 and 10 with 10 being the most unbearable pain can also be treated as interval cause you are letting the person choose any number between two anchor points. These kinds of values should be differentiated from  the ‘range’ of code values of ordinal data as these values represent specific categories where each of the numbers have a value label assigned to them (e.g., 1= very dissatisfied, 2= dissatisfied, 3= neutral, 4= satisfied, 5=very satisfied).  In ordinal data there is no assurance that the distance between the feelings being categorized as numbers is equal.

There's a huge debate ongoing in the social / behavioral sciences about Likert type variables. Whether individual Likert items can be considered as interval-level data, or whether they should be considered merely ordered-categorical data is the subject of disagreement. Many regard such items only as ordinal data, because, especially when using only five levels, one cannot assume that respondents perceive all pairs of adjacent levels as equidistant. On the other hand, often the wording of response levels clearly implies a symmetry of response levels about a middle category; at the very least, such an item would fall between ordinal- and interval-level measurement; to treat it as merely ordinal would lose information. Further, if the item is accompanied by a visual analog scale, where equal spacing of response levels is clearly indicated, the argument for treating it as interval-level data is even stronger. Most people will treat individual items of the scale (each statement) as an ordinal level response. However, when the response to each of the items are added together to create a scaled score, this total score is then usually treated as interval level.

As you continue through this course you will see how vitally important it is to understand the level of measurement. Knowing the kind of measurement level your variables are helps you decide what statistical test to run
(view example).

Required Readings/ Videos

• View the following video about levels of measurement (12:30) :
  http://www.youtube.com/watch?v=B0ABvLa_u88

Learning Activity

• Complete the Learning Activity found here.
• Check your answers here.