TOPIC 32
SAMPLE SIZE IN QUANTITATIVE
STUDIES
“How many participants will I need?” The answer is not as straightforward as it may seem.
Sample size is one of the many decisions that researchers must make and be able to sup-
port with logic. Major factors that influence the quantitative researcher’s choice of sample
size include the purpose of the study, the population size from which the sample is being
drawn, the amount of variability within the population, the risk of selecting a sample that
is a poor representation of the population, and the amount of precision wanted from the
results.
Sample size expectations differ depending on the goal of the study. For instance,
researchers frequently conduct pilot studies, which are studies designed to obtain pre-
liminary information on how new treatments and measures work. For instance, when
studying a new drug, a researcher might conduct a pilot study on the maximum tolerable
dose. Also, a researcher might try out a new measure, such as a questionnaire, to determine
if any questions are ambiguous, or if there are questions that participants refuse to answer.
Pilot studies are usually conducted with small sample sizes, such as 10 to 100. On the basis
of the results, treatments and measures can be modified for use in more definitive future
studies with larger samples.
If the research plan is to analyze data using statistics, a probability sample—otherwise
known as a random sample—is needed. Topics 26 through 28 covered the main ways that
probability samples are drawn. As students and new researchers learn to use statistics, they
will find that sample size relates to mathematical properties. Remember that statistics are
a type of analysis that approximates information from a sample to a known population.
Sample size is determined by identifying the minimum number of people or units needed
to represent that population well and to produce useful results when subjected to statistical
analysis.
To offer a good approximation of the population, a sample needs to be large enough
to adequately represent the population and its variability in the relevant areas of inquiry.
Most studies have constraints on time, effort, and money that can be spent to complete
data collection. Collecting more than the minimum number needed does not hurt a study,
but it may also do very little to help create more accurate results. This is because, as more
people are added, the effect of each person is a bit less to the total.
To illustrate this principle, consider the following two cases in Table 32.1, in which the
sample size was increased by 50 participants.
It is clear that Researcher A gets a much bigger payoff for increasing the sample size by
50 because it doubles the sample size, which gives a big boost in precision. Researcher B,
on the other hand, gets a very small increase in precision because the addition of 50 to the
original sample of 3,000 has little influence on the results. The responses of the 50 addi-
tional participants will be overwhelmed by the responses of the first 3,000.
SAMPLE SIZE IN QUANTITATIVE STUDIES 111
TABLE 32.1 Illustrates Increasing Sample Size
Original Add to sample New Sample Size Added sample
Sample Size portion of
population
Researcher A 50 +50 100 50%
Researcher B 3,000 +50 3,050 1.6%
Population size can affect the appropriate sample size, especially if the population is
under 20,000. When the population is larger than 20,000, the recommended sample size
does not change much based on population size. Because of this principle of diminishing
returns, even the most important and prestigious national surveys are often conducted with
only about 1,500 respondents. This book’s authors reviewed sample sizes across articles in
five journals published by the American Psychological Association and found that sample
sizes ranged from 25 to 766, with a median of 82.1 It is a good idea to review the literature
in your area of study to see the sample size ranges of studies similar to your own.
Table 32.2 provides some helpful estimates for sample sizes in survey research that has
a goal of description. The recommended sample sizes (n) correspond to a population size
(N) that has an error of 5%. That is, the true percentage in the whole population should
fall within 5% of the percentage obtained from the sample.
Notice that until the population size is over 400, the sample size is over half of the total.
This further illustrates a practical application of the principle of diminishing returns. The
recommendations in Table 32.2 are based on a few simple assumptions and should not be
accepted as a rule but rather a guideline. If samples sizes are smaller than those listed, it
may still be possible to conduct statistical analyses, but the margin of error will be larger.
There are many sample size calculators online that allow the user to determine some basic
parameters, including population size and margin of error, as well as determine the right
sample size.2
Important considerations for determining sample size include variability, the risk of
selecting a sample that is a poor representation of the population, and the precision that
the researcher wants. Researchers can obtain accurate results from a small sample if a
TABLE 32.2 Recommended Sample Sizes (n) for Populations (N) with
Finite Sizes3
N n N n N n N n
10 10 150 108 750 254 10,000 370
25 24 200 130 1000 278 20,000 377
50 44 300 169 1500 306 50,000 381
75 63 400 196 2000 322 75,000 382
100 80 500 217 5000 357 100,000 384
112 SAMPLING
population has very little variability (i.e., the population is homogeneous) and a normal
distribution (values revolve around a central, most common value; more in Topic 61).
For instance, the population of healthy babies born at 40 weeks is homogeneous and
converges toward a consistent average. Human babies only have limited variation at birth,
thankfully. Studying a random sample of only 100 babies is likely to give a researcher
about the same information as studying a random sample of 1,000.
By contrast, if a population is very heterogeneous, a researcher needs a large sample.
Suppose, for instance, a researcher wants to study the use of healthcare in the last six
months among adults in a large metropolitan area. There undoubtedly is great variability,
ranging from no use to extensive use. If the researcher draws a small sample of 50 individ-
uals from such a diverse population, the researcher might obtain a result with major errors,
such as not having a single individual with extensive use of healthcare in the sample. If
the researcher also wants to categorize those in the sample by the type of health insurance
they have, an even larger sample will be needed to capture a sample that represents all the
categories in the population.
When researchers want to document something that is rare, such as incidence of heart
attacks among men aged 18 to 30, a larger sample is needed because it will occur to so few
individuals. If a researcher drew a small sample, such as 25 participants, and followed them
for a year, it is quite likely that no heart attacks would occur, but it would be a mistake to
conclude that no heart attacks occur in this age group. Studies of this type might rely on
datasets from very large samples that were collected by others in order to have an adequate
sample size.
When researchers are looking for small differences, they also need larger sample sizes
to minimize the risk of drawing a nonrepresentative sample. Suppose the population of
women voters is just slightly more in favor of a ballot proposition than is the population
of men voters. In small random samples of these voters, the risk of selecting a nonrep-
resentative sample is high. Sampling errors could easily overwhelm the small difference
and might even produce a result in which it appears that the men are slightly more in
favor of the proposition than are women. Using large samples in this case is necessary if a
researcher wants to identify the correct direction and size of such a small difference.
■ TOPIC REVIEW
1. Does the purpose of the study matter for sample size?
2. True or false: “Once you know the population size, the sample size is an easy calculation.”
3. What are pilot studies?
4. Do researchers usually use small or large samples in pilot studies?
5. If a researcher suspects that a trait is rare in a population, should he or she use a small
or a large sample to identify the incidence of the trait?
6. Suppose a researcher suspects that there is only a very small difference in the math
abilities of boys and girls at the sixth-grade level. Should the researcher use a small or
a large sample to measure this difference?
7. Suppose the population consists of church members in the Southern Baptist Conven-
tion. If a researcher believes the members of this population are very homogeneous in
their belief in an afterlife, should the researcher use a small or a large sample to identify
the percentage who hold such a belief?
SAMPLE SIZE IN QUANTITATIVE STUDIES 113
8. What do N and n stand for in Table 32.2?
9. If the population size is 5,000, what is the recommended sample size?
10. At what population size does the sample size stop changing considerably?
■ DISCUSSION QUESTION
1. Some research articles based on small samples in academic journals are characterized
by their authors as pilot studies. What justifies the publication of pilot studies? Why
might they be rejected?
■ RESEARCH PLANNING
If you are planning a quantitative study, what is your anticipated sample size based on the
information in this topic and the topics in this part thus far? Explain the basis for your
decision. What factors affect your sample size considerations?
■ NOTES
1. The journals are Journal of Counseling Psychology, Journal of Educational Psychology, Cultural Diversity and Ethnic
Minority Psychology, American Journal of Orthopsychiatry, and Professional Psychology: Research and Practice.
2. Examples of sample size calculators intended for research purposes are www.raosoft.com/samplesize.
html and www.surveysystem.com/sscalc.htm. Online survey software may also offer a similar guideline or
calculator—just be sure that the guidelines are intended for your type of research or discipline.
3. Adapted from a combination of Krejcie, R.V., & Morgan, D.W. (1970). Determining sample size for
research activities. Educational and Psychological Measurement, 30, 607–610, and the use of the Raosoft sample
size calculator with 50% response distribution for 5% margin of error.
TOPIC 33
SAMPLE SIZE AND DIVERSITY IN
QUALITATIVE RESEARCH
Qualitative researchers are not collecting data that will be analyzed using statistical meth-
ods, so the mathematical and probabilistic rules used to determine quantitative sample sizes
and selection do not have the same relevance. Topic 29 discussed differences in sampling
strategies. In qualitative research, sampling is still tied to analytical methods, but is more
reliant on the judgment of the researcher. As a result of this difference, qualitative research
has often been subject to criticism for being less rigorous. However, qualitative researchers
often respond that the two types of research have different logics and base the evaluation
of rigor on different characteristics.
Qualitative data is collected in a number of ways: through interviews, observation,
focus groups, or other modes of rich data collection in which the data from each subject
is more extensive than quantitative data. Qualitative methods allow researchers to collect
local knowledge and give voice to participants. Qualitative researchers are often interested
in gathering data that is in a context. Context here means information about things such
as the particular organization, geographic location, or period of time that may contribute
to understanding the responses of the people included in the study.
Qualitative researchers are not trying to generalize to a known population, but still may
be interested in exploring a topic that adds knowledge about a process or a concept. Often,
a first question is whether the research focuses on capturing information within one case
or will include more than one case, comparing the data for similarities or differences.
Addressing this decision focuses the researcher on the theory, concept, process, or other
underlying aspects of the research, allowing the researcher to refine the research question.
For instance, a professor might want to understand why so few women choose to
major in engineering. For such a study, sampling women only from one campus would
be appropriate. However, it might also be valuable to seek participants from multiple
campuses. The researcher might want to find out about the differences in the size and
composition of the campuses from which the women were selected and compare women’s
responses based on these differences as one way of gaining understanding of their choices.
The use of participants from diverse sources is a methodological strength of a qualitative
study when the researcher has a broader interest than just a single source (such as a single
college campus). Example 1 illustrates how one group of researchers sought participants
from diverse sources for a qualitative study.
Example 1
To begin the sampling procedure, 15 well-regarded mental health clinicians with a
minimum of 5 years of professional experience were chosen as interviewees. These
clinicians were chosen because of their: (a) involvement in therapeutic services to
clients with developmental disabilities and severe and persistent mental illnesses;
SAMPLE SIZE AND DIVERSITY IN RESEARCH 115
(b) long-standing involvement with clients’ interdisciplinary teams working with
parents; and (c) noted consistent excellence of work in the field serving the DD popu-
lation, both in treating clients and training employees and direct support service staff.
Each key informant was asked to nominate two professionals: licensed social workers,
group home coordinators or supervisors, or licensed registered nurses with case loads
of at least 20 clients within the large metropolitan area of the initial sample group.1
The researchers in Example 1 also formulated criteria for the nominations of additional
professionals, which included level of experience and the excellence of their work. Many
qualitative researchers develop samples by placing themselves in a context and coming to
know those involved, or using experts and key informants to reach a larger group. Another
way is to recruit from a selected setting and create ways to allow those in the setting to opt
in to the study. Example 2 is a qualitative study that explores what role family plays when
African Americans choose to pursue doctoral degrees. Semi-structured interviews were
conducted with 41 African American students who were currently enrolled as doctoral
students in any discipline at one of two research-intensive programs—one at a private uni-
versity and one at a public university. The researchers explained their sampling approach
as follows:
Example 2
I sent emails to doctoral program coordinators at both institutions requesting they
forward an announcement to students soliciting participants. I also attended a gradu-
ate student association meeting at each campus. A total of 30 participants responded.
Snowball sampling was used to recruit the remaining participants (Stage & Manning,
2003).2
Snowball sampling, discussed in Topic 29, is a form of sampling in which people who are
already participating can refer other appropriate participants. Often, multiple forms of
sample recruitment are utilized as was the case in both examples. This is fairly common
because different methods may work best at different phases of data collection.
Qualitative researchers usually use fewer participants than quantitative researchers.
The literature on qualitative research methods points out that few guidelines exist for
determining sample sizes for nonprobabilistic samples.3 The literature on sample size
for qualitative research varies in recommendations for a minimum sample size, usually
between 20 and 50 participants. The authors reviewed sample sizes for qualitative research
articles in five journals published by the American Psychological Association and found
that qualitative research studies ranged from 10 to 26 participants with an average of 13.4
To be published, sociological studies generally require a slightly larger sample size—20
or more.
Because of the inductive nature of much qualitative research, the exact sample size may
not be determined at the outset of the study but will rely on the analysis that takes place as
more data is collected. Researchers often rely on the concept of saturation, first covered
in Topic 29. To recap, saturation relies on simultaneous analysis, in which major and minor
themes of the research emerge. When adding several participants fails to yield new infor-
mation that leads to the identification of additional themes or categories within a theme,
116 SAMPLING
the researcher might conclude that the data collection process has become saturated. Satu-
ration suggests that new types of information are unlikely to be obtained by increasing the
sample size. Thus, the point of saturation may determine the final sample size.
Most sources will say that sample sizes rely on the type of research, the topic, and the
judgment of the researcher. The sampling strategy also has an effect. Deviant case sam-
pling may be good with as few as one case, and homogeneous sampling strategies may start
to find saturation in 10 cases. Maximum variation sampling is a frequently used purposive
sampling strategy, and typically requires the largest minimum samples because of its goal
of determining that variability has been exhausted.5 This is often determined by getting
redundant answers.
■ TOPIC REVIEW
1. Do qualitative samples rely on rules of probability for determining the appropriate
size?
2. Are qualitative researchers always interested in problems that extend beyond one loca-
tion or institution?
3. The use of participants from diverse sources is a methodological strength of a qualitative
study when the researcher has what?
4. How might a qualitative researcher obtain participants for a sample?
5. On the basis of this topic, what was the average range of sample sizes for qualitative
studies?
6. At what point might a qualitative researcher conclude that the data collection process
has become saturated?
■ DISCUSSION QUESTION
1. Qualitative methods have been subject to many criticisms aimed at the sampling
methods used. What questions or concerns do you have with qualitative sampling
methods? What strengths are related to qualitative approaches to research?
■ RESEARCH PLANNING
If you are planning to conduct qualitative research, which approach do you think you will
take to planning your sampling strategy? Even if you plan to collect data based on a
model of reaching saturation, do you have an estimated range for the sample size you
think your research project will require? Consider whether you will collect from more
than one site or from people in more than one role.
■ NOTES
1. Van Ingen, D.J., Moore, L.L., & Fuemmeler, J.A. (2008). Parental overinvolvement: A qualitative study.
Journal of Development & Physical Disabilities, 20, 449–465.
SAMPLE SIZE AND DIVERSITY IN RESEARCH 117
2. McCallum, C.M. (2016). ‘Mom made me do it’: The role of family in African Americans’ decisions to
enroll in doctoral education. Journal of Diversity in Higher Education, 9, 50–63.
3. Gues, G., Bunce, A., & Johnson, L. (2006). How many interviews are enough? An experiment with data
saturation and variability. Field Methods, 18, 59–82. doi:10.1177/1525822X05279903
4. The journals are Journal of Counseling Psychology, Journal of Educational Psychology, Cultural Diversity and Ethnic
Minority Psychology, American Journal of Orthopsychiatry, and Professional Psychology: Research and Practice.
5. Sandelowski, M. (1995). Sample size in qualitative research. Research in Nursing & Health, 18, 179–183.
