Descriptive Statistics: Describing Variables and the Relations among Them

● The levels of a nominal scale are simply different categories or groups that have no intrinsic
numerical properties. For example, two different kinds of therapies for depression would be
nominal categories. Variables using an ordinal scale rank order the levels from lowest to highest
(or least to most), but the intervals between each rank order are not equal. A list of the top ten
restaurants in Halifax would use an ordinal scale. For an interval scale, the distances between
each level are equivalent in size.In contrast, ratio scales have equal intervals in addition to a true
zero point. Response time and age are examples of ratio scales.
● The statistical procedures used to analyze data with interval and ratio variables are identical.
Importantly, data measured on interval and ratio scales can be summarized using an arithmetic
average, or what is known as the mean.
● The mean is used in many statistical analyses. Variables measured on interval and ratio scales are
often referred to as continuous variables because the values represent an underlying continuum
Distributions
● A frequency distribution indicates the number of participants who receive or select each possible
score on a variable. Frequency distributions can be created for variables using any scale of
measurement
● any outliers: scores that are unusual, unexpected, impossible, or very different from the scores of
other participants. An outlier might reflect a data-entry error that can be corrected
Bar Graphs
● A bar graph uses a separate and distinct bar for each piece of information. Bar graphs are
commonly used for comparing group means but can also be used for comparing percentages
Pie Charts
● A pie chart divides a whole circle, or “pie,” into “slices'' that represent relative percentages. Pie
charts are particularly useful when representing data on a nominal scale. Instead of using a bar
graph for the data we could have divided a circle into five sections corresponding to the five
response options.
Histograms
● A histogram uses bars to display a frequency distribution for a continuous variable. In this case,
the values along the x-axis are continuous and show increasing amounts of a variable, such as
blood pressure, reaction time, or number of correct response
● a normal distribution, a distribution of scores that is frequently observed, and rather important for
statistics. In a normal distribution, the majority of the scores cluster around the mean.This
distribution is only possible for continuous variables
● (SD) is a common measure of variability, or how the scores are spread out with respect to the
mean
Frequency Polygons
● an alternative to histograms, use a line to represent frequencies for continuous variables (i.e.,
interval or ratio scales). Frequency polygons are especially helpful when you want to examine
frequencies for multiple groups simultaneously
Descriptive Statistics
● main types of descriptive statistics are (1) measures of central tendency and (2) measures of
variability. Measures of central tendency try to capture how participants scored overall, across the
entire sample, in various ways. In contrast, measures of variability attempt to summarize how
 differently the scores are from each other, or how widely the scores are spread out or distributed.
Both types of descriptive statistics help us to summarize the scores for a variable
Central Tendency
● A central tendency statistic tells us what the scores are like as a whole, or how people scored on
average. There are three measures of central tendency: (1) the mean, (2) the median, and (3) the
mode.
Variability
● A measure of variability characterizes the amount of spread in a distribution of scores, for
continuous variables (i.e., interval or ratio scales). One common measure of variability is the
standard deviation, which indicates how far away the scores tend to be from the mean, on
average. In scientific reports, it is abbreviated as SD, and in formulas, it is abbreviated as s.The
standard deviation is derived by first calculating the variance, symbolized as s 2 .n, the range is
the difference between the highest score (i.e., the maximum; Max.) and the lowest score
Comparing Groups of Participant:
Comparing Group Percentages-calculate the percentage for each group
Comparing Group Means-Many studies are designed to compare mean responses to continuous variables
made by participants in two or more nominal groups
Graphing Nominal Data
● A common way to graph relationships between variables when one variable is nominal is to use a
bar graph or a line graph.The levels of the nominal variable (in this case, Australian and Mexican)
are represented on the horizontal x-axis, and the dependent variable values are shown on the
vertical y-axis. For each group, a point is placed along the y-axis that represents the mean for the
groups, and a bar is drawn to visually represent the mean value.Bar graphs are often used when
the values on the x-axis are nominal categories .In contrast, line graphs are used when the values
on the x-axis are numeric
Describing Effect-Size between Two Groups
● Effect-size is a general term for these indicators and is a vital descriptive tool to help us interpret
how large our effects are. Effect-size can be measured in many different ways, depending on the
study design. When comparing two groups (e.g., two conditions in an experiment, two
nationalities) on their responses to a continuous variable, one appropriate effect-size is Cohen’s d
● Cohen’s d is the difference in means between two groups, standardized by expressing it in units
of standard deviation.Cohen’s d value describes the magnitude of the effect of the independent
variable on the dependent variable. In a study comparing naturally occurring groups, the Cohen’s
d value describes the magnitude of the effect of group membership on a continuous variable
Describing Relationships among Continuous Variables: Correlating Two Variables
● Pearson correlation coefficient, which is usually just called the
Pearson correlation, or Pearson r. This is the correlation we use when
both variables have interval or ratio-scale properties.The nearer a
correlation is to 1.00 (positive or negative), the stronger the
relationship. In fact, a +1.00 or –1.00 correlation is sometimes called
a perfect relationship, because changes in the two variables follow
one another in a perfect fashion. The sign of the Pearson r tells us
about the direction of the relationship, whether it is a positive or a
negative relationship. A correlation coefficient of −.74 indicates a
 negative relationship that is stronger than the positive relationship
indicated by a coefficient of +.21.
Scatterplots
● A scatter plot is a type of data visualization that displays individual data points on a two-
dimensional graph. Each point on the graph represents the values of two variables, one plotted on
the horizontal axis (x-axis) and the other on the vertical axis (y-axis). Scatter plots are particularly
useful for examining the relationship between two continuous variables
Important Considerations
● Restriction of range-It is important that the researcher sample from the full range of possible
scores for both variables. If the full range is not sampled, but instead this range is restricted (i.e.,
narrowed), the correlation coefficient produced with these data can be misleading.The problem of
restriction of range can occur when the people in your sample are all very similar on one or both
of the variables you are studying.
● Curvilinear Relationship -The Pearson correlation is designed to detect only linear relationships.
If the relationship is not linear but curvilinear, for example ), the correlation coefficient will fail
to detect this relationship.Because the variables in your data might have a curvilinear relationship,
it is important to always inspect a scatterplot of the data before looking at the magnitude of the
correlation. The scatterplot is valuable because it provides a visual indication of the shape of the
relationship.
Correlation Coefficients as Effect-Sizes
● Effect-size estimates provide a scale of values that is consistent and can be compared across
studies, regardless of the variables, the particular design, or the number of participants.
Correlation coefficients not only allow us to examine relationships between continuous variables,
they are also indicators of effect-size.the range of r 2 values run from 0.00 (0 percent) to 1.00
(100 percent). This r 2 value, or squared correlation coefficient, is sometimes referred to as the
amount of shared variance between the two variables
The Regression Equation
● Like a correlation, regression analyzes relationships among variables. In fact, the calculations for
correlation and regression result in the same value when there are only two variables involved.
However, using a regression framework can be more powerful because it allows us to expand the
analysis to include more variables. Analyzing data using regression requires calculating a
regression equation, which is the same as an equation for drawing a straight line. But this is not
just any line: It’s the line that best summarizes all of the data points.Y = a + bX
● Y is the score we wish to predict (called the criterion variable), X is the known score (called the
predictor variable), a is the y-intercept (a constant, where the line hits the y-axis or the value of y
when x = 0), and b is the slope of the line (a weighting adjustment factor that is multiplied by X).
Multiple Correlation and Multiple Regression
● A technique called multiple correlation (symbolized as R, to distinguish it from Pearson r)
provides the correlation between a combined set of predictor variables and a single criterion
variable.e multiple correlation usually generates a stronger correlation than the single correlation
between any one of the predictor variables and the criterion variable.s, R 2 tells you the
proportion of variability in the criterion variable that is accounted for by the combined set of
predictor variables.
 ● This expanded model is often called multiple regression. This technique allows us to examine the
unique relationship between each predictor and the criterion. This is in contrast to multiple
correlation, which only provides a single value for the relationship between the combined set of
predictors and the criterion variable
● Is one variable a better predictor of life satisfaction than the other? That question must be
addressed using multiple regression. Using the two predictors of life satisfaction
Partial Correlation and the Third-Variable Problem
● partial correlation—helps to address a specific issue: the third-variable problem. The third-
variable problem occurs when two variables are correlated, but we don’t know if some third
variable might be the reason they are related . Partial correlation provides a way of statistically
controlling for possible third variables in correlational analyses. The result of a partial correlation
is a correlation between the two variables of interest, with the influence of a third variable
controlled, the original correlation. It estimates what the correlation between the two primary
variables would be if the third variable
● To calculate a partial correlation, you need to have scores on the two primary variables of interest
as well as the third variable that you want to control for.he researcher suspects that a third
variable may explain this association: satisfaction with income. Greater satisfaction with income
could increase both life satisfaction and the sense that one has freedom of choice in one’s life.