pp: 119-130
By
Abstract
This paper describes the procedure for determining sufficient sample size for both continuous
and categorical data. Tables for determining the sufficient sample size were given as adapted
from Krejcie and Morgan (1970) and Bartlett, et al., (2001). The main objective of this study is
to explore the procedures for determining a sufficient sample size for management survey
research activities. The study recommends that researchers should endeavour to use sufficient
sample size in conducting survey studies to minimize errors in inference back to the
population.
Introduction
1
In order to solve managerial problems, managers engage in research. The purpose of research is
to collect data that will help management solve problem that have been identified and defined.
Data can be collected in a variety of ways. One of the most common methods is through the use
of surveys. Surveys can be done by using a variety of methods. The most common methods are
the telephone survey, the mailed questionnaire survey, personal interview, surveying records, and
direct observation (Bluman, 2004).
Management research reports require reliable forms of evidence from which to draw robust
conclusions. It is usually not cost effective or practicable to collect and examine all the data that
might be available. Instead it is often necessary to draw a sample of information from the whole
population to enable the detailed examination required to take place. Samples can be drawn for
several reasons: for example to draw inferences across the entire population; or to draw
illustrative examples of certain types of behavior. Bluman (2004) argues that in conducting the
research, management use samples to collect data and information about a particular variable
from a large population. Using samples saves time and money and, some cases enable the
researcher to get more detailed information about a particular subject. A simple survey of
published manuscripts reveals numerous errors and questionable approaches to sample size
2
ii)
Sample Design
Bluman (2004) argues samples can not be selected in a haphazard way because the information
obtained might be biased. To obtained samples that are unbiased (i.e. give each subject of the
population an equally likely chance of being selected) management use four basic methods of
sampling: random, systematic, stratified, and cluster sampling.
Sample design covers the method of selection, the sample structure and plans for analyzing and
interpreting the results. Sample designs can vary from simple to complex and depend on the type
of information required and the way the sample is selected. The design will impact upon the size
of the sample and the way in which analysis is carried out. In simple terms the tighter the
required precision and the more complex the design, the larger the sample size.
Standard textbook authors and researchers offer tested methods that allow studies to take full
advantage of statistical measurements, which in turn give researchers the upper hand in
determining the correct sample size. Sample size is one of the four inter-related features of a
study design that can influence the detection of significant differences, relationships or
interactions (Peers, 1996). Generally, these survey designs try to minimize both alpha error
(finding a difference that does not actually exist in the population) and beta error (failing to find
a difference that actually exists in the population) (Peers, 1996).
Alpha Level
In many educational and social research studies, it is not feasible to use any of the first three
ways and the researcher must estimate variance using the fourth method (Bartlett, 2001). A
researcher typically needs to estimate the variance of scaled and categorical variables. To
estimate the variance of a scaled variable, one must determine the inclusive range of the scale,
and then divide by the number of standard deviations that would include all possible values in
the range, and then square this number.
When estimating the variance of a dichotomous (proportional) variable such as gender, Krejcie
and Morgan (1970) recommended that researchers should use .50 as an estimate of the
population proportion. This proportion will result in the maximization of variance, which will
also produce the maximum sample size. This proportion can be used to estimate variance in the
population. For example, squaring .50 will result in a population variance estimate of .25 for a
dichotomous variable.
Cochrans (1977) correction formula should be used to calculate the final sample size:
no
n1 = -----------------------------(1 + no / P)
Where P = population size
Where n0 = required return sample size according to Cochrans formula
Where n1 = required return sample size because sample exceed required % of population.
9
However, since many educational and social research studies often use data collection methods
such as surveys and other voluntary participation methods, the response rates are typically well
below 100%. Salkind (1997:107) recommended oversampling when he stated that If you are
mailing out surveys or questionnaires, . . . . count on increasing your sample size by 40%-50% to
account for lost mail and uncooperative subjects.
However, many researchers criticize the use of over-sampling to ensure that this minimum
sample size is achieved and suggestions on how to secure the minimal sample size are scarce.
Fink (1995) argues that Oversampling can add costs to the survey and is often necessary.
If the researcher decides to use oversampling, Bartlett, et al., (2001) suggest that four methods
may be used to determine the anticipated response rate: (1) take the sample in two steps, and use
the results of the first step to estimate how many additional responses may be expected from the
second step; (2) use pilot study results; (3) use responses rates from previous studies of the same
or a similar population; or (4) estimate the response rate.
Estimating response rates is not an exact science. A researcher may be able to consult other
researchers or review the research literature in similar fields to determine the response rates that
have been achieved with similar and, if necessary, dissimilar populations (Bartlett, et al., 2001).
Categorical Data
10
Cochrans (1977) correction formula should be used to calculate the final sample size. These
calculations are as follows:
no
n1= -----------------------------(1 + no / Population)
Where no = required return sample size according to Cochrans formula
Where n1 = required return sample size because sample > required % of population.
(These procedures result in a minimum returned sample size).
Methodology
11
A crude method of checking the sufficiency of data is describe as split-half analysis of consistency
(Martin & Bateson, 1986). Here the data is divided randomly into two halves which are then analysed
separately. If both sets of data clearly generate the same conclusions, then sufficient data is claimed to
have been collected. If the two conclusions differ, then more data is required. True split half analysis
involves calculating the correlation between the two data sets. If the correlation coefficient is sufficiently
high then the data can be said to be reliable (Hill, 1998). Martin and Bateson (1986) advocate a
correlation coefficient of greater than 0.7.
12
A formula for determining sample size can be derived provided the investigator is prepared to specify
how much error is acceptable and how much confidence is required (Roscoe, 1975; and Alreck & Settle,
1995 in Hill, 1998).
A significance level of 5% has been established as a generally acceptable level of confidence in most
behavioral sciences (Hill, 1998). Roscoe (1975) seems to use 10% as a rule thumb acceptable level, while
Wersberg and Bowen (1977) cite 3% and 4% as the acceptable level in survey research for forecasting
election results. Further, Wersberg and Bowen (1977: 41) provide a table of maximum sampling error
related to sample size for simple randomly selected samples. The table insinuates that if you are prepared
to accept an error level of 5% in survey, then you require a sample size of 400 observations. If 10% is
acceptable then a sample of 100 is acceptable, provided the sampling procedure is simple random.
Error
2,000
2.2
1,500
2.6
1000
3.2
750
3.6
700
3.8
13
4.1
500
4.5
400
5.0
300
5.8
200
7.2
100
10.3
14
N-----n
N-----n
N-----n
N-----n
10-----10
100-----80
280-----162
800-----260
2800-----338
15-----14
110-----86
290-----165
850-----265
3000-----341
20-----19
120-----92
300-----169
900-----269
3500-----346
25-----24
130-----97
320-----175
950-----274
4000-----351
30-----28
140-----103
340-----181
1000-----278
4500-----354
35-----32
150-----108
360-----186
1100-----285
5000-----357
40-----36
160-----113
380-----191
1200-----291
6000-----361
45-----40
170-----118
400-----196
1300-----297
7000-----364
50-----44
180-----123
420-----201
1400-----302
8000-----367
55-----48
190-----127
440-----205
1500-----306
9000-----368
60-----52
200-----132
460-----210
1600-----310
10000-----370
65-----56
210-----136
480-----214
1700-----313
15000-----375
70-----59
220-----140
500-----217
1800-----317
20000-----377
75-----63
230-----144
550-----226
1900-----320
30000-----379
80-----66
240-----148
600-----234
2000-----322
40000-----380
85-----70
250-----152
650-----242
2200-----327
50000-----381
90-----73
260-----155
700-----248
2400-----331
75000-----382
95-----76
270-----159
750-----254
2600-----335
100000-----384
Krejcie and Morgan (1970) state that, using the above calculation, as the population increases the
sample size increases at a diminishing rate and remains, eventually constant at slightly more than
380 cases.
15
Table 3: Table for determining minimum returned sample size for a Given Population Size for
Continuous and categorical Data
Sample size
17
Continuous data
(margin of error=.05)
(margin of error=.03)
Population size
Alpha = .10, alpha = .05, alpha = .01, alpha = .50, alpha = .50, alpha =.05,
t=1.65
t= 1.96
t= 2.58
t=1.65
t= 1.96
t=2.58
100
46
55
68
74
80
87
200
59
75
102
116
132
154
300
65
85
123
143
169
207
400
69
92
137
162
196
250
500
72
96
147
176
218
286
600
73
100
155
187
235
316
700
75
102
161
196
249
341
800
76
104
166
203
260
363
900
76
105
170
209
270
382
1,000
77
106
173
213
278
399
1,500
79
110
183
230
306
461
2,000
83
112
189
239
232
499
4,000
83
119
198
254
351
570
6,000
83
119
209
259
362
598
8,000
83
119
209
262
367
613
10,000
83
119
209
264
370
623
Accordingly, researchers may use this table if the margin of error shown is appropriate for their
study, however, the appropriate sample size must be calculated if these error rates are not
appropriate (Bartlett, et al., 2001).
18
Conclusions
Although it is not unusual for researchers to have different opinions as to how sample size should
be calculated, the procedures used in the process should always be reported, allowing the reader
to make his or her own judgments as to whether they accept the researchers assumptions and
procedures. In general, a researcher should use standard factors in determining the sample size.
Using adequate sample along with high quality data collection efforts will results in more
reliable, valid, and generalizable results, it could also result in time saving and other resources
(Bartlett, et al., 2001).
From the foregoing, it appears that determining sample size is not a cut-and-dried procedure.
Also the nature of methodology used is a major consideration in selecting sample size (Hill,
1998). Due to problems arising from getting enough respondents and generalization of findings,
Gay and Diehl (1992) opined that there is a great deal in the replication of findings. Hill (1998)
suggests replication of findings to mean: (i) increase the subject pool, and (ii) create greater
validity for generalizability.
Recommendation
Having describe the procedures involved in determining sufficient sample size for management
survey research activities, the following recommendations are proffer:
19
References
Alreck, P.L. and Settle, R.B. (1995). The Survey Research Handbook. (2nd.). in Hill, R. (1998).
What Sample Size is Enough in Internet Survey Research. Interpersonal
Computing and Technology: An electronic Journal for the 21 st Century. Available at:
http://www.emoderators.com/ipct-j/1998/n3-4/hill.hmtl
Bartlett, J.E., Kotrlik, J.W., and Higgins, C.C. (2001). Organizational Research: Determining
Appropriate sample Size in Survey Research. Information Technology, Learning, and
Performance Journal, 19(1), pp 43-50 (Spring).
20
22