Invitation to Psychology

34 Chapter 1 What Is Psychology?

Further, p values can vary considerably from one

replication to another, which is why many findings

don’t pan out in later studies (Miller, 2011).

To protect against spurious results, many jour-

nals now encourage or require the use of alternate

methods that yield an effect size, which helps us

understand how important an effect is. One such

measure reveals how much of the variation in the

data is accounted for by the independent variable.

If it explains 5 percent of the variation, it’s not

too meaningful, even if the result is statistically

significant; if it explains 40 percent, it’s impressive.

Another useful set of statistical techniques,

called meta-analysis, combines and analyzes data

from a large number of related studies instead of as-

sessing each study’s results separately. Meta-analysis

tells the researcher how powerful an independent

variable is across all the studies in the analysis.

Suppose we did 10 studies on everybody’s

favorite subject, gender differences. We might get

contradictory results, or some results that were

significant and others that were not. Meta-analysis

can come to the rescue, providing us with a clearer

picture. For example, what is the reason for the

effect size An objec-
tive, standardized way of
describing the strength
of the independent vari-
able’s influence on the
dependent variable.

meta-analysis A set of
techniques for combin-
ing and analyzing data
from a number of related
studies to determine the
explanatory strength of
a particular independent
variable.

gender gap in math achievement, which persists

in some nations but not others? Is it due mainly to

a “natural” male superiority in math, or to gender

differences in educational and professional op-

portunities? A meta-analysis of studies across 69

nations, representing nearly 500,000 students ages

14 to 16, found that although boys have more posi-

tive attitudes toward math than girls, average effect

sizes in actual mathematics achievement are small.

However, national effect sizes show considerable

variability; that is, a male-female math gap is wider

in some countries than others. The strongest pre-

dictors of that cross-national variation are whether

boys and girls are equally likely to be enrolled in

school, the percentage of women in research jobs,

and women’s representation in their nation’s gov-

ernment (Else-Quest, Hyde, & Linn, 2010).

Techniques such as meta-analysis are useful

because rarely does one study prove anything, in

psychology or any other field. That is why you

should be suspicious of headlines that announce

a sudden major scientific breakthrough based on

a single study. Such breakthroughs do occur, but

they are rare.

Recite & Review

Recite: Major scientific breakthrough: Reciting what you’ve read really helps! So go ahead and re-

cite everything you know about the arithmetic mean, the standard deviation, statistical significance,

confidence intervals, cross-sectional versus longitudinal studies, effect sizes, and meta-analysis.

Review: Next, go back and reread this section.

Now take this Quick Quiz—but not while you’re driving or multitasking!

A. Check your understanding of the descriptive–inferential distinction by mentally placing a check

in the appropriate column for each phrase:

Study and Review at mypsychlab

Scientific progress C. meta-analysis B. inferential 6. inferential 5. inferential 4. descriptive 3. inferential 2. descriptive1. A.

usually proceeds gradually, not all at once. And besides, anyone can post a claim on the Internet, so you would want to

ask: What is the original source of that claim? What was the science supporting it? How was shyness defined, and what

was the method of treating it? How was success defined and measured?

Descriptive Statistics Inferential Statistics

1. Summarize the data


2. Give likelihood of data occurring by chance


3. Include the mean


4. Give a measure of statistical significance


5. Tell you whether to call your mother about
   your results


6. Provide a confidence interval around the
   sample mean
   B. Suppose a researcher wants to know how energy drinks affect memory, but in the 45 stud-
   ies on this question the results have varied. What method might help determine whether the
   relationship between energy drinks and memory is strong or weak?
   C. On the Internet, you read about a “Fantastic New Finding” about treating shyness. Why should
   you be cautious about this announcement?
   Answers: