Volume 14 Issue 1 Article 3 
4-24-2012 
A perspective on scale development in entrepreneurship research 
Alenka Slavec 
Mateja Drnovšek 
Follow this and additional works at: https://www.ebrjournal.net/home 
Recommended Citation 
Slavec, A., & Drnovšek, M. (2012). A perspective on scale development in entrepreneurship research. 
Economic and Business Review, 14(1). https://doi.org/10.15458/2335-4216.1203 
This Original Article is brought to you for free and open access by Economic and Business Review. It has been 
accepted for inclusion in Economic and Business Review by an authorized editor of Economic and Business 
Review. 
39      ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  2012  |  39–62
A PERSPECTIVE ON SCALE DEVELOPMENT 
IN ENTREPRENEURShIP RESEARCh
aLEnka SLavEc1 
MatEJa dRnovŠEk2 
abStRact: To develop a measurement scale that would lead to valid and reliable results 
is a tasking goal in any research field. We build from theoretical findings that are advanced 
in the measurement and scale development theory to review the type of measures used and 
scale development procedures of measures reported in top-tier entrepreneurship journals 
during the years 2009 and 2010. The in-depth review showed that not all steps in scale 
development in entrepreneurship research are always accomplished although every step is 
crucial when aiming for a sound measure. Additionally, the study showed that all too often 
measures, that have not been validated, are being used. Thus, this review serves as a refer-
ence for researchers to apply methodologically rigorous procedures for scale development 
in entrepreneurship research and to use previously validated scales. With this in mind, we 
propose hands on guidelines for scale development in entrepreneurship research in the Ap-
pendix of the paper.
key words: scale development, measurement procedure, reliability, validity, entrepreneurship, construct
JEL classification: L26
1.  intRoduction
To develop a measurement scale that would lead to valid and reliable results is a chal-
lenging task in any research field. It takes time and reflection. Several scholars argue 
that effective measurement is a cornerstone of scientific research (e.g. DeVellis, 2003; 
Netemeyer, Bearden, & Sharma, 2003) and is a central component of good assessment of 
latent variables (C. R. Reynolds, 2010) and empirical research (Crook, Shook, Madden, 
& Morris, 2009). Besides that, reliable and valid measures contribute to the legitimacy 
and development of a research field. Additionally, C. R. Reynolds (2010) stated that the 
ability and skill to measure variables accurately is a cornerstone to progress in science. 
Nevertheless, empirical articles that use rigorous methodological procedures, besides 
being firmly grounded in theory, receive more citations and are more important to the 
development of science (Colquitt & Zapata-Phelan, 2007). The field of entrepreneurship 
is no exception (kuskova, Podsakoff, & Podsakoff, 2011).
The progress of the measurement issue in the scientific discipline of entrepreneurship 
has been investigated by several scholars (e.g. Chandler & Lyon, 2001; Crook et al., 
1 University of Ljubljana, Faculty of Economics, Slovenia, Ljubljana, e-mail alenka.slavec@ef.uni-lj.si
2 University of Ljubljana, Faculty of Economics, Slovenia, Ljubljana, e-mail mateja.drnovsek@ef.uni-lj.si
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201240
2009; Crook, Shook, Morris, & Madden, 2010; Dean, Shook, & Payne, 2007; kuskova et 
al., 2011; Low, 2001; Low & MacMillan, 1988). They found that as a young and rapidly 
growing field (Dean et al., 2007; Shane & Venkataraman, 2000) entrepreneurship has 
evidenced improvements in research design and constructs measurement (Crook et al., 
2010). Despite the methodological progress, we are not there yet (Crook et al., 2010). The 
main criticism is that the field has not undertaken empirical research with the same rigor 
as other fields (Low, 2001) and extant research relies far too heavily on measures that do 
not allow for or report on the assessment of reliability, validity, model fit, measurement 
error (Crook et al., 2009; Crook et al., 2010) and other important issues in construct de-
velopment like content domain specification, item pool generation, and pilot study con-
duction. Still, the appropriate use of and sufficient sophistication in the use of analytical 
techniques is a critical component to the advancement and legitimacy of the entrepre-
neurship field (Dean et al., 2007). While research design and construct measurement in 
entrepreneurship research have already been evaluated, a study that would assess scale 
development approaches in the scale development process for the population of entre-
preneurship papers has not been undertaken yet. 
With this study we aim to contribute towards more rigorous scale development in en-
trepreneurship research by investigating crucial steps in the scale development proce-
dure and pointing out specific problems associated with each step. Additionally, in the 
Appendix of the paper we provide hands on guidelines for developing new measures 
for researchers to get an overview of the crucial steps that should be undertaken when 
developing new measures. Moreover, we also aim at spurring the awareness that vali-
dated scales should be used when testing constructs. The motivation for this research 
springs from authors’ systematic review of studies published and observation that in 
entrepreneurship journals many times measures have not been properly conceptualized, 
operationalized or applied. When proposing a new measure some important steps in the 
scaling procedure have been overlooked. Of concern is also the fact that all too often not-
validated measures are extensively used. Both represent a problem since researchers in 
subsequent studies may use these measures and results may be questionable. Although 
we try to critically evaluate scaling procedures reported in papers, we note our debt to 
previous researchers who performed the exploratory work upon which more recent re-
search is built (Chandler & Lyon, 2001; Low, 2001; Low & MacMillan, 1988). By assessing 
the methodological part of previous works we aim to identify critical deficiencies in scal-
ing procedures and propose alternatives to overcome them with guidelines in the form of 
a ten-step methodology that we attach in the Appendix of this paper. We partially adopt 
Low and MacMillaǹ s (1988), Chandler and Lyoǹ s (2001), and Crook et al.̀ s (2010) re-
search setting and evaluate contributions and shortcomings of the methodological part 
of the research design specification. In contrast to their lens of view, we focus on the type 
of measures applied and the steps in scale development reported in the reviewed papers. 
These findings provide an admonishment for future research and in particular for scale 
development in entrepreneurship research. 
In what follows, we discuss and evaluate the measures used and research setting of 
scale development applied in entrepreneurship research. Our database includes a sam-
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 41
ple of 197 papers published in 2 top-tier entrepreneurship journals (Journal of Business 
Venturing and Entrepreneurship Theory and Practice) from january 2009 to December 
2010. In so doing we contribute to the entrepreneurship discipline by evaluating the 
current state of the type of measures used and scale development procedures applied 
in entrepreneurship research and by identifying problematic issues in the measure-
ment part of entrepreneurship papers. To overcome the problematic issues we provide 
an overview of the crucial steps in scale development in the form of a ten-step three-
phased methodology in the Appendix of the paper and offer implications for future 
research.
2.  REviEw oF MEaSuRES in EntREpREnEuRShip JouRnaLS
In our study we partly followed Low and MacMillaǹ s (1988), Chandler and Lyoǹ s 
(2001), and Crook et al.̀ s (2010) framework of reviewing papers in entrepreneurship 
journals. The starting point of our study was to take inventory of scale development 
procedures and type of measures reported. Specifically, we addressed the following two 
questions. (1) Are measures that are used verified, i.e. referenced? (2) If not, what was 
the procedure for developing new measures? For this analysis we included papers pub-
lished from january 2009 to December 2010 in the following two journals: Journal of 
Business Venturing and Entrepreneurship Theory & Practice. The selection of journals 
was based on the ranking scale of SSCI journals in entrepreneurship research as on 
October 2010; we analyzed the first two ranked entrepreneurship journals on this scale. 
The selection of journals is consistent with other research settings in entrepreneurship 
(e.g. Crook et al., 2009; Crook et al., 2010; Dean et al., 2007). In Journal of Business 
Venturing we reviewed Volume 24, Issue 1-6 and Volume 25, Issue 1-6 and in Entrepre-
neurship Theory & Practice we reviewed Volume 33, Issue 1-6 and Volume 34, Issue 1-6. 
In 2009 and 2010 a total of 197 papers were published in the two reviewed journals. The 
detailed results on the number of empirical papers and measures reviewed by journal 
and year are presented in Table 1 in section 2.2 Results. This time span seems reasonable 
to get an appropriate review of the current state of the measurement issue in entrepre-
neurship research.
2.1 Method
The content analysis of the 197 papers was conducted as follows. First, each paper was 
categorized as either empirical or conceptual. For a paper to be categorized as empirical 
it had to include some kind of data or data analysis (Chandler & Lyon, 2001). A category 
of “conceptual paper” was assigned to those papers that included literature reviews, un-
tested theoretical models, mathematical models (Chandler & Lyon, 2001), conceptual 
papers, case studies, teaching cases and review papers. Studies that both presented and 
tested theory with empirical data were categorized as empirical studies (Chandler & 
Lyon, 2001). When a paper was categorized as conceptual, the review ended at this stage 
while empirical papers were further assessed.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201242
Empirical papers were analyzed using a rating form for measures developed specifically 
for this study. Measures were coded as dummy (0,1) variables; we either found evidence 
supporting the categorization (1) or we did not (0) (Chandler & Lyon, 2001). For each 
measure we analyzed whether it was correctly referenced (1) or not (0). A referenced 
measure is a measure that has been previously used by other researchers and studies. 
Further, some measures were categorized as adapted measures, i.e. the ones that have a 
reference and have been slightly modified from its original form and authors reported 
on that. In the rating form in the appropriate column an adapted measure was noted 
down as a 1; if this was not the case we wrote a 0. For those measures that did not have 
a reference, we analyzed whether they were developed throughout the paper and two 
additional types of measures were identified at this stage. The first is a neither referenced 
nor developed measure. This is a measure that neither has references of previous uses (i.e. 
does not report on previous uses of the measure) nor is developed throughout the study 
(i.e. does not report on scale development procedures). If a measure was qualified as a 
neither referenced nor developed measure it got a 1, if this was not the case we wrote a 0. 
The second is a new measure which is a measure that is developed in the study, reports 
on some or all steps in scale development procedures and has not been previously used. 
Again, if a measure was categorized as a new measure, we noted down a 1, otherwise it 
got a 0. Finally, some measures were categorized as measures that do not need a reference. 
A measure of this type is qualified as the one that does not need to be validated or to be 
referenced. Examples for such measures are: gender, age, education, marital status, race, 
industry classification, parental entrepreneurship, firm legal form, etc. A single measure 
was categorized only as one type of measure (e.g. a referenced measure) and could not be 
categorized as several types of measures (e.g. not a referenced measure and a new meas-
ure at the same time). Figure 1 summarizes the categorization of the measures, whereas 
Table 2 in section 2.2 Results reports the results on the categorization of measures.
Figure 1: Categorization of measures
For measures that were categorized as new measures we investigated the scale develop-
ment procedure that was reported in papers. Based on the extant literature regarding 
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 43
scale development (e.g. Bagozzi & Edwards, 1998; Carmines & Zeller, 1979; Churchill, 
1979; DeVellis, 2003; hinkin, 1998; Netemeyer et al., 2003; Nunnally & Bernstein, 1994; 
Pedhazur & Pedhazur Schmelkin, 1991) we identified ten steps that were crucial for 
developing new measures for entrepreneurship-related studies. If a step was reported in 
the paper, we denoted it in the rating form with a 1. We noted down a 0 if the step has 
not been reported. The ten steps were grouped into three phases and are presented in 
Figure 2.
Figure 2: Ten steps and three phases in scale development
The ten scale development steps are grouped into three phases. The first phase regards 
the theoretical importance and existence of the construct, the second phase deals with 
the representativeness and appropriateness of data collection, and the third phase re-
gards the statistical analysis and statistical evidence of the construct. The ten steps 
that comprise the three phases were the following (for more detailed descriptions of 
the ten steps see the Appendix). The first step regarded the content domain specifica-
tion since the development of a new measure starts with outlining the domain of the 
new construct. It is achieved by an in-depth literature review (Netemeyer et al., 2003). 
This step consists of stating what the researcher means by the use of a particular word 
by providing a definition of the new construct (Nunnally & Bernstein, 1994). If the 
domain of the new construct was specified, we noted down a 1 in the rating form. If 
this was not the case, a 0 was marked. In the second step a researcher generates a pool 
of potential items that sample the domain of the construct. From this pool of items the 
new scale will be derived. In the third step we investigated whether the content validity 
has been evaluated. Content validity refers to the adequacy of sampling the domain of 
the construct (Nunnally & Bernstein, 1994), i.e. the assessment of the adequacy of the 
proposed items from the item pool by the relevant audience. The first three steps repre-
sent the first phase of scale development, i.e. the theoretical importance and existence 
of the construct.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201244
The second phase on the representativeness and appropriateness of data collection is 
comprised by the next four steps (steps from 4 to 7). The fourth and fifth steps are option-
al since they are dependent on the type of research setting. If the data collecting method 
was a survey, the forth step investigated whether the papers reported on questionnaire 
development and evaluation (1) or not (0). The fifth step is mandatory for cross-national 
studies where different languages are spoken. here, the translation and back-translation 
is required and we investigated whether papers reported on translation and back-trans-
lation (1) or not (0).
The sixth step investigated whether a pilot study was performed (1) or not (0). This 
step is recommendable and useful to test the proposed questionnaire and measure. 
In the pilot study a researcher identifies potential problems with the questionnaire 
(Dillman, Smyth, & Christian, 2009) and obtains a pre-result of the reliability of the 
new measure and correlations among items. In the seventh step we analyzed whether 
sampling and data collection was reported (1) or not (0). A representative sample is 
crucial for the quality of results. Not much less important is the procedure of collect-
ing data.
In the final three steps which comprise the third phase on the statistical analysis and 
statistical evidence of the construct, we investigated whether dimensionality (1), reliabil-
ity (1), and construct validity (1) were assessed or not (0). A measurè s dimensionality is 
concerned with the homogeneity of items (Netemeyer et al., 2003). Reliability concerns 
the extent to which a measurement procedure yields the same results on repeated trials 
(Carmines & Zeller, 1979). Construct validity is a scalè s ability to measure what it is 
supposed to measure (haynes, Richard, & kubany, 1995; hinkin, 1998). Despite some 
disagreements about the classification of different types of validity, two main types of 
construct validity should be evaluated – convergent and discriminant validity. Conver-
gent validity refers to the degree to which different measures of the same construct are 
correlated (Netemeyer et al., 2003). Discriminant validity assesses the degree to which 
two measures design to measure similar, but conceptually different constructs, are relat-
ed (Netemeyer et al., 2003). If papers reported on other types of validity these additional 
validities were noted down.
2.2 Results
In Volume 24 and Volume 25 in Journal of Business Venturing and in Volume 33 and 
Volume 34 in Entrepreneurship Theory & Practice a total of 197 papers were published. 
Among 197 papers, 98 papers were classified as empirical and 99 as conceptual. In the 
98 empirical papers 1171 measures were reported. The number of empirical papers and 
measures reviewed per journal per year is shown in Table 1.
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 45
Table 1: Number of empirical papers and measures reviewed by journal and year
Journal/Year 2009 2010 Total
# of 
papers
# of
measures
# of 
papers
# of
measures
# of 
papers
# of
measures
Journal of Business 
Venturing
21 205 25 399 46   604
Entrepreneurship Theory & 
Practice
31 328 21 239 52   567
Total 52 533 46 638 98 1171
Most of the empirical papers reported that several measures have been used in their 
studies and a total of 1171 measures have been identified. From the 1171 measures, 370 
(31.6%) measures were categorized as measures that do not need a reference. The total 
number of correctly referenced measures was 299 (25.5%), while the total number of 
adapted measures was 42 (3.6%). The most critical category of measures is the one that 
has measures that neither have a reference nor are developed within the study. A total of 
383 (32.7%) measures were of this kind. Finally, 77 (6.6%) measures were new measures 
that have been developed for the purpose of the study. The summary of the results is 
presented in Table 2. 
Table 2: Number and percentage of categorized measures 
Type of measure
Total number of
measures
Percentage of
measures (in %)
A measure that does not need a reference   370   31.6
Referenced measure   299   25.5
Adapted measure     42     3.6
Neither referenced nor developed measure   383   32.7
New measure     77     6.6
Total 1171 100.0
Most of the measures that were categorized as measures that do not need a reference were 
control variables and/or with obvious structures, e.g. gender, age in number of years, 
highest level of education achieved, industry with SIC classification, whether a business 
is a family business or not, etc. Some adapted measures reported on reliability and some 
also on other steps in scale development, but these measures did not categorize as newly 
developed measures. Our analysis revealed that almost a third of measures (32.7%) nei-
ther had a reference of previous uses nor were developed for the purpose of their research. 
These measures are regarded as problematic since they cannot be considered reliable, 
valid and effective. There are three main reasons for this thesis. The first is that if no 
previous research used these measures, we cannot know if they are good measures and 
confidence in the results of such studies is limited (Crook et al., 2010). While we are 
aware that researchers might have used referenced measures (and not neither referenced 
nor developed measure) or that steps in scale development procedure have been under-
taken, we were unable to control for these shortcomings if papers did not report on that. 
Secondly, these measures also did not go through steps in scale development which are 
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201246
essential for the evaluation of the effectiveness of a measure. Finally, subsequent studies 
may use these measures and results might be biased.
The 77 new measures were evaluated in accordance to the ten steps in scale development. 
The results of the study show that a large proportion of new measures overlooked some 
steps in scale development. The summary of the number and percentage of measures that 
accomplished each of the ten steps is presented in Table 3. In what follows we describe 
these results in details.
Table 3: Summary of scale development steps reported in papers that developed new 
measures
Step in scale development
Number of measures 
that adopted specific 
steps in scale 
development
Percentage of measures 
(in %) that adopted 
specific steps in scale 
development
(1) content domain specification 13 16.9
(2) item pool generation 19 24.7
(3) content validity evaluation 13 16.9
(4) questionnaire development and evaluation 13 16.9
(5) translation and back-translation of the questionnaire 11 14.3
(6) pilot study performance 8 10.4
(7) sampling and data collection 13 16.9
(8) dimensionality assessment 20 26.0
(9) reliability assessment 70 90.9
(10) construct validity assessment 28 36.4
Most of the papers reported the reliability assessment of the new measure (90.9%). Con-
struct validity was assessed in 36.4% of cases. 26.0% of measures have been evaluated 
for their dimensionality, while item pool generation has been reported in 24.7% of cases. 
Other steps have been accomplished in less than 20% of cases.
While scale development begins with the specification of the domain, i.e. what is the 
construct about and what is not to be included in the construct, and with proposing a 
definition of the new construct (Nunnally & Bernstein, 1994), only 16.9% of measures 
have accomplished the first step. It is important to note that a clearly specified content 
domain is the basis for all subsequent steps (Netemeyer et al., 2003). Moreover, a lot of 
measures have omitted the item pool generation step. These studies reported only which 
items have been used although these items may not be the most appropriate ones. Accord-
ingly, C. R. Reynolds (2010) emphasizes that only proposing a limited number of items 
and beginning research on a new construct is not appropriate. In this way a researcher 
ignores crucial initial steps in scale development (C. R. Reynolds, 2010). Of great concern 
is the fact that only 16.9% of measures have been evaluated for their content validity, i.e. 
items have been administered to the relevant audience like entrepreneurship scholars, 
entrepreneurs, entrepreneurship students, experts from entrepreneurship and psycho-
metric specialists to test if the domain has been sampled correctly and if instructions and 
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 47
formats are appropriate and clear. While at this stage suggestions by judges are made for 
additional inclusion of items or for their eventual exclusion and modification, most of 
the papers did not report on that.
Step 4 investigated whether papers reported on questionnaire development, evaluation and 
refinement. While this is not a mandatory step for studies that collect data in other ways, 
16.9% of new measures have reported on the questionnaire issue. Additionally, Step 5 is 
mandatory only for cross-cultural research where different languages are used. 16.9% of 
measures have reported on translation and back-translation of items or questionnaires. 
Before evaluating the new measure on a large sample, it is suggested to conduct a prelim-
inary study on a moderately sized sample on the targeted population (Clark & Watson, 
1995; Netemeyer et al., 2003). Eight measures have undergone this step. The sampling 
and data collection step has a low percentage (16.9%) since sampling and data collection 
was done for the purpose of the whole study and mostly not for the purpose of testing the 
measure before conducting the study as a whole.
Finally, the reliability assessment step was accomplished in 90.9% of cases. Although 
the result is acceptable, it is not yet satisfactory. Accordingly, measures can be reliable 
without being valid, but cannot be valid without being reliable (Chandler & Lyon, 2001). 
The study revealed that only 26.0% and 36.4% of measures have been evaluated for di-
mensionality and construct validity, respectively. But there is consensus that no measure 
is useful in the long run without evidence of its validity (Nunnally & Bernstein, 1994).
Overall, there are two main shortcomings in regard to the measurement issue in en-
trepreneurship research that spring out of our study. First, we noted that all too often, 
specifically in 32,7% of cases, inappropriate scales to measure constructs were used and 
this refers to the neither referenced nor developed measures. This for sure represents a 
relatively high portion of measures for an emerging but growing field that tries to get 
legitimacy. Despite that there is also an encouraging portion of measures that are cor-
rectly referenced and applied and this includes referenced and adapted measures in the 
amount of 29,1% of all measures reported. Secondly, the scale development review seems 
even more challenging since the majority of steps have not been accomplished to a high 
percentage. The most reassuring result is that researchers are aware that reliability of a 
measure is really important. Despite that, much work should be done in future research 
to bring also other important steps in scale development to a satisfactory accomplish-
ment; beginning from specifying what is the measure about, generating a large pool of 
items and evaluating the measure for its content validity, that is considering and accom-
plishing the first phase () on which all the subsequent stages in scale development lie. 
3.  concLuSion
With this study we made a further step towards raising the awareness that the methodo-
logical rigor in establishing new measures and using previously validated measures is a 
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201248
crucial predisposition for the development of any research field, and in particular also 
of the young and growing field of entrepreneurship. By reviewing the type of measures 
used and the procedures applied when developing new measures in papers published 
from january 2009 to December 2010 in Journal of Business Venturing and Entrepreneur-
ship Theory & Practice, we acknowledged two main problematic issues. Among the 1171 
measures reported in 98 empirical papers from our study (1) one third (32.7%) of the 
measures reviewed were neither developed nor previously used and cannot be counted 
as appropriate since they were neither referenced nor developed measures and (2) when 
developing new measures several steps were omitted and again such new measures are 
problematic. The main argument is that no matter how profound the theoretical formu-
lations, how sophisticated the design, and how elegant the analytic techniques are, they 
cannot compensate for poor measures (Pedhazur & Pedhazur Schmelkin, 1991). Simi-
larly, judge et al. (2007) opinion that regardless of the quality of an idea, the ability to 
draw inferences about a phenomenon is constrained by the quality of the methods used 
to gather data about it.
Thus, inappropriate measures and poor measurement development threaten the un-
derstanding of entrepreneurship phenomena and paradigm development. Accord-
ingly, several scholars agree that appropriate measurement with valid and reliable 
measures is the foundation of scientific research and progress in science (e.g. Colquitt 
& Zapata-Phelan, 2007; Crook et al., 2009; DeVellis, 2003; judge et al., 2007; kuskova 
et al., 2011; Netemeyer et al., 2003; C. R. Reynolds, 2010) while acknowledging the 
importance of theoretical development (Colquitt & Zapata-Phelan, 2007; Dean et al., 
2007). In addition to that, research findings can be called into question if the under-
lying constructs were not measured properly implying that construct measurement 
is the foundation of quality empirical research (kerlinger & Lee, 2000). Crook et al. 
(2010) add that to the extent that a study is not properly designed and when constructs 
are not properly measured, not only is confidence in the findings of that study lim-
ited but also the field’s ability to build on it in future studies. For this reason there 
is growing accordance that empirical studies that are firmly grounded in theory and 
use valid measures of their focal constructs are more critical to the advancement of 
science than studies that do not possess these attributes (judge et al., 2007; kuskova 
et al., 2011)
The importance of applying rigorous methodology in general, and specifically in regards 
to measure use or measure development, is strengthened by studies that analyzed the 
influence of measurement rigor on citation count (Bergh, Perry, & hanke, 2006; Colquitt 
& Zapata-Phelan, 2007; judge et al., 2007; kuskova et al., 2011). For example, in their 
study, Colquitt and Zapata-Phelan (2007) found that articles which rated (on their rat-
ing instrument) as moderate to high on theory building and theory testing enjoyed the 
highest level of citations. Since the number of citation a paper will get is also predicted 
by the quality of the journal in which the paper is published (meaning that a high journal 
average citation rate has a positive influence on the number of citations a paper will get 
(judge et al., 2007)), researchers should put great effort into sound measurement devel-
opment to publish in top-tier journals and get cited more.
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 49
To the extent that entrepreneurship is a young and developing research field (Dean et al., 
2007; Shane & Venkataraman, 2000) with methodological rigorousness to be improved 
(Crook et al., 2010; Low, 2001) it is not superfluous to emphasize also with the results 
of our study that sustained efforts should be undertaken to reach an adequate level of 
methodological rigorous. All the more is this true for the young discipline of entrepre-
neurship since it has been argued that relatively new disciplines tend to have less clearly 
defined theoretical and methodological paradigms than more mature disciplines (e.g. 
kuskova et al., 2011; Pfeffer, 1993). And both, theory building and theory testing repre-
sent key components of theoretical contribution that can coexist within a given stream 
of research (Colquitt & Zapata-Phelan, 2007).
Based on the findings of our study we can conclude, similarly to Crook et al (Crook et al., 
2010), that we are not there yet. Therefore, it is important that researchers use validated 
scales or to develop new scales according to an established procedure. To get an overview 
of the crucial steps in scale development in entrepreneurship research we propose a ten-
step methodology in the Appendix of this paper. Although there is no definitive checklist 
and that procedures vary among authors, the ten steps summarize all crucial steps for 
developing new measures in entrepreneurship research. While the ten steps are in part 
tailored for entrepreneurship research, they are consistent with the main-stream recom-
mendations for developing new measures (e.g. Bagozzi & Edwards, 1998; Carmines & 
Zeller, 1979; Churchill, 1979; DeVellis, 2003; hinkin, 1998; Netemeyer et al., 2003; Nun-
nally & Bernstein, 1994; Pedhazur & Pedhazur Schmelkin, 1991) and for the interested 
readers we suggest consulting also those readings. 
In short, the ten-step procedure that is grouped into three phases is the following. The 
first phase deals with the theoretical importance and existence of the construct and 
is comprised by the first three steps. The first step in developing a new construct is to 
specify the content domain of the construct and to propose its definition. In the second 
step an item pool that samples the domain of the new construct is generated. The third 
step regards the evaluation of content validity by assessing the relevance of items by the 
relevant audience, e.g. entrepreneurs, entrepreneurship experts, and scholars. here, the 
second phase on the representativeness and appropriateness of data collection begins 
and is comprised by steps 4 to 7. In the fourth step the questionnaire is developed and 
evaluated. The fifth step is mandatory for entrepreneurial cross-cultural studies in differ-
ent languages since the translation and back-translation of the questionnaire is needed. 
In the sixth step a pilot study is conducted on a relevant audience, presumably entrepre-
neurs. Then, in the seventh step random sampling on a dataset of entrepreneurs and data 
collection occur with established tailored design methods for improving response rates. 
The final three steps represent the third phase on the statistical analysis and statistical 
evidence of the construct and involve assessment of dimensionality, reliability, and con-
struct validity of the new entrepreneurship-related measure.
We hope these guidelines will contribute to a better and more effective approach for 
introducing new constructs in entrepreneurship domain and to the advancement of the 
entrepreneurship field. In so doing, quality measurement design will reach the rigor of 
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201250
empirical research as it is in other fields that several scholars of entrepreneurship called 
for (e.g. Chandler & Lyon, 2001; Crook et al., 2009; Crook et al., 2010; Dean et al., 2007; 
Low, 2001; Low & MacMillan, 1988).
3.1. Limitations and future research avenues
There are two main limitations of our study. The first concerns the limited number of 
journals reviewed. It might be that reviewing a larger variety of journals may lead to 
slightly different results. Despite this, we reviewed the two highest-quality entrepreneur-
ship journals and it is quite likely that reviewing lesser quality journals (Crook et al., 2010) 
would reveal major measurement problems. Accordingly, it has been shown that higher 
quality journals tend to publish articles that possess more rigorous methods (kuskova et 
al., 2011). Secondly, it might be the case that researchers used referenced measures (and 
not neither referenced nor developed measure) or that steps in scale development of new 
measures have been accomplished but the papers did not report on that. Unfortunately, 
we were not able to control for these shortcomings. This leads us to stressing the im-
portance of consistent reporting of scale development steps, measures that were applied 
in studies and on other measurement-related steps. Therefore, a further implication at-
tained from our study is that researchers in the field of entrepreneurship need to provide 
more precise descriptions of measures used or descriptions on the operationalization of 
their new measures. All in all we call for a valid use of established measures and rigorous 
development of new measures.
REFEREncES
Bagozzi, R. P., & Edwards, j. (1998). A general approach for representing constructs in organizational re-
search. Organizational Research Methods, 1 (1), 45-87.
Bagozzi, R. P., Yi, Y., & Phillips, L. W. (1991). Assessing construct validity in organizational research. Admin-
istrative Science Quarterly, 36 (3), 421.
Bergh, D. D., Perry, j., & hanke, R. (2006). Some predictors of SMj article impact. Strategic Management 
Journal, 27 (1), 81-100.
Brislin, R. W. (1970). Back-translation for cross-cultural research. Journal of Cross-Cultural Psychology, 1 (3), 
185-216.
Campbell, D. T., & Fiske, D. W. (1959). Convergent and discriminant validation by the multitrait-multimeth-
od matrix. Psychological Bulletin, 56 (2), 81-105.
Carmines, E. g., & Zeller, R. A. (1979). Reliability and validity assessment (Vol. 17). Newbury Park, CA: Sage 
Publications, Inc.
Chandler, g. N., & Lyon, D. W. (2001). Issues of research design and construct measurement in entrepreneur-
ship research: The past decade. Entrepreneurship: Theory & Practice, 25 (4), 101-113.
Churchill, g. A. j. (1979). A paradigm for developing better measures of marketing constructs. Journal of 
Marketing Research, 16 (2), 64-73.
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 51
Clark, L. A., & Watson, D. (1995). Constructing validity: Basic issues in objective scale development. Psycho-
logical Assessment, 7 (3), 309-319.
Collins, D. (2003). Pretesting survey instruments: An overview of cognitive methods. Quality of Life Re-
search, 12 (3), 229.
Colquitt, j. A., & Zapata-Phelan, C. P. (2007). Trends in theory building and theory testing: A five-
decade study of the Academy of Management journal. Academy of Management Journal, 50 (6), 1281-
1303.
Cronbach, L. j. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16 (3), 297-334.
Crook, T. R. et al. (2009). A review of current construct measurement in entrepreneurship. International 
Entrepreneurship and Management Journal, 6 (4), 387-398.
Crook, T. R. et al. (2010). Are we there yet? An assessment of research design and construct measurement 
practices in entrepreneurship research. Organizational Research Methods, 13 (1), 192-206.
Crowne, D. P., & Marlowe, D. (1960). A new scale of social desirability independent of psychopathology. Jou-
mal of Consulting Psychology, 24, 349-354.
Dean, M. A., Shook, C. L., & Payne, g. T. (2007). The past, present, and future of entrepreneurship research: 
Data analytic trends and training. Entrepreneurship: Theory & Practice, 31 (4), 601-618.
DeVellis, R. F. (2003). Scale development: theory and applications (2nd ed. Vol. 26). Thousand Oaks, CA: Sage 
Publications.
Dillman, D. A., Smyth, j. D., & Christian, L. M. (2009). Internet, mail, and mixed-mode surveys: the tailored 
design method (3rd ed.). hoboken, N.j.: john Wiley & Sons.
Flynn, B. B., Schroeder, R. g., & Sakakibara, S. (1994). A framework for quality management research and an 
associated measurement instrument. Journal of Operations Management, 11 (4), 339-575.
hair, j. F. et al. (2010). Multivariate data analysis (7th ed.). Upper Saddle River, Nj: Prentice-hall.
hardesty, D. M., & Bearden, W. O. (2004). The use of expert judges in scale development: Implications 
for improving face validity of measures of unobservable constructs. Journal of Business Research, 57 (2), 
98-107.
haynes, S. N., Richard, D. C. S., & kubany, E. S. (1995). Content validity in psychological assessment: A func-
tional approach to concepts and methods. Psychological Assessment, 7 (3), 238-247.
hinkin, T. R. (1998). A brief tutorial on the development of measures for use in survey questionnaires. Or-
ganizational Research Methods, 1 (1), 104-121.
hui, C. h., & Triandis, h. C. (1985). Measurement in cross-cultural psychology: A review and comparison of 
strategies. Journal of Cross-Cultural Psychology, 16 (2), 131-152.
jackson, D. N. (1971). The dynamics of structured personality tests. Psychological Review, 78 (3), 229-248.
judge, T. A. et al. (2007). What causes a management article to be cited—Article, author, or journal? . Acad-
emy of Management Journal, 50 (3), 491-506.
kerlinger, F. N., & Lee, h. B. (2000). Foundations of behavioral research (4th ed.). Fort Worth, TX: harcourt 
College Publishers.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201252
kuskova, V. V., Podsakoff, N. P., & Podsakoff, P. M. (2011). Effects of theoretical contribution, methodological 
rigor, and journal quality, on the impact of scale development articles in the field of entrepreneurship. Strate-
gic Entrepreneurship Journal, 5 (1), 10-36.
Leavitt, C., & Walton, j. (1975). Development of a scale for innovativeness. Advances in Consumer Research, 
2, 545-554.
Loevinger, j. (1957). Objecive tests as instruments of psychological theory. Psychological Reports, 3, 635-694.
Low, M. B. (2001). The adolescence of entrepreneurship research: Specification of purpose. Entrepreneurship: 
Theory & Practice, 25 (4), 17-25.
Low, M. B., & MacMillan, I. C. (1988). Entrepreneurship: Past research and future challenges. Journal of 
Management, 14 (2), 139-161.
Netemeyer, R. g., Bearden, W. O., & Sharma, S. (2003). Scaling procedures: Issues and applications. Thousand 
Oaks, CA: Sage Publications.
Nunnally, j. C., & Bernstein, I. h. (1994). Psychometric theory (3rd ed.). New York: Mcgraw hill, Inc.
Pedhazur, E. j., & Pedhazur Schmelkin, L. (1991). Measurement, design, and analysis: An integrated approach. 
hillsdale, New jersey: Lawrence Erlbaum Associates Inc., Publishers.
Pfeffer, j. (1993). Barriers to the advance of organizational science: Paradigm development as a dependent 
variable. The Academy of Management Review, 18 (4), 599-620.
Presser, S., & Blair, j. (1994). Survey pretesting: Do different methods produce different results? Sociological 
Methodology, 24, 73-104.
Presser, S. et al. (2004). Methods for testing and evaluating survey questions. The Public Opinion Quarterly, 
68 (1), 109-130.
Reynolds, C. R. (2010). Measurement and assessment: An editorial view. Psychological Assessment, 22 (1), 
1-4.
Reynolds, W. M. (1982). Development of reliable and valid short forms of the Marlowe-Crowne social desir-
ability scale. Journal of Clinical Psychology, 38 (1), 119-125.
Schwab, D. P. (1980). Construct validity in organization behavior. In Staw, B. M. & Cummings, L. L. (Eds.), 
Research in Organizational Behavior (Vol. 2, pp. 3-43). greenwich, CT: jAI Press Inc.
Shane, S., & Venkataraman, S. (2000). The promise of entrepreneurship as a field of research. Academy of 
Management Review, 25 (1), 217-226.
Strahan, R., & gerbasi, k. C. (1972). Short, homogeneous versions of the Marlow-Crowne social desirability 
scale. Journal of Clinical Psychology, 28 (2), 191-193.
Zaichkowsky, j. L. (1985). Measuring the involvement construct. Journal of Consumer Research, 12 (3), 341-
352.
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 53
appEndiX
guidelines for scale development in entrepreneurship research: A ten-step procedure
To strengthen the measurement part of entrepreneurship-related papers we propose 
a ten-step procedure for scale development in entrepreneurship research. While our 
guidelines are tailored for entrepreneurship research, they are consistent with the main-
stream recommendations for developing new measures (e.g. Bagozzi & Edwards, 1998; 
Carmines & Zeller, 1979; Churchill, 1979; DeVellis, 2003; hinkin, 1998; Netemeyer et al., 
2003; Nunnally & Bernstein, 1994; Pedhazur & Pedhazur Schmelkin, 1991). We grouped 
the ten steps of scale development into three phases: (1) theoretical importance and ex-
istence of the construct, (2) representativeness and appropriateness of data collection, 
and (3) statistical analysis and statistical evidence of the construct. The three phases with 
the ten steps are discussed in what follows.
Phase 1: Theoretical importance and existence of the construct
The first phase regards the theoretical importance and existence of the construct and is 
comprised by the following three steps: (1) content domain specification, (2) item pool 
generation, and (3) content validity evaluation. Figure 3 represents the scale development 
procedure with the first phase marked in different color.
Figure 3: Phase 1 – Theoretical importance and existence of the construct
Step 1: Content domain specification
The first step in constructing a new measure is to clearly define what the researcher in-
tends to measure (DeVellis, 2003). Accordingly, in the entrepreneurial context it should 
be clearly stated what is the new entrepreneurial construct about, what is to be included 
in the construct and not less importantly also what is to be excluded from the construct. 
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201254
Netemeyer et al. (2003) explain that it is crucial to assess the boundaries of the construct 
domain and not to include extraneous factors or domains of other constructs. Addition-
ally, the domain of the construct should not be set to narrowly and fail to include impor-
tant facets of the construct. The researcher should also explain how the new construct is 
different from other related construct that already exist. At this stage an initial definition 
of the new construct should be proposed (Nunnally & Bernstein, 1994). This working 
definition might be modified during the scale development process. Besides the defini-
tion, it is important to specify dimensions of the new construct (haynes et al., 1995). 
Consistent with scale development experts (e.g. DeVellis, 2003; Netemeyer et al., 2003), 
we propose that this step is to be achieved by an in-depth interdisciplinary literature 
review. The review of the relevant literature serves as the basis for grounding the theory 
of the new construct. The review should include previous attempts to conceptualize and 
assess both the same construct and closely related constructs (Clark & Watson, 1995). 
Additionally, we suggest that in entrepreneurship research the existence of the new con-
struct should be explored also in practice. At the very beginning of scale development, 
interviews with the relevant audience from entrepreneurship should be conducted. We 
suggest conducting semi-structured interviews on the topic of the target construct with 
entrepreneurs, entrepreneurship professors, business angels, venture capitalists, repre-
sentatives of technology parks, business incubators, chambers of trade, and financial 
institutions that deal with entrepreneurs and the like. These interviews should evidence 
the importance and the need for the new measure. Interviews should be taped, tran-
scribed, and analyzed in order to get feedback on eventual inclusion of some aspects of 
construct domain that might have not been included. Similarly, suggestions on exclu-
sion of some other parts of the domain might be raised. We also propose conducting a 
focus group with 5-7 entrepreneurial experts, practitioners or other relevant profiles that 
would brainstorm, confront and upgrade ideas and views on the target construct. Like 
the interviews, the focus group should be taped, transcribed, and analyzed. Interviews 
and focus groups can be effectuated in several rounds if needed. After each round, rel-
evant literature should be reviewed again for parts that might have been missing. In each 
round, an improved definition is proposed and interviews become more structured. A 
researcher should also analyze interviews, web pages, statements and lectures of entre-
preneurial scholars, entrepreneurs, venture capitalists, business angels, etc. and consult 
popularly-scientific media on the target topic to gather additional insights.
At the final stage of content domain specification, the researcher should be able to clearly 
state why the new measure is needed given that several entrepreneurial constructs exist. 
The explanation should contain evidence that the older measures, if they exist, do not 
satisfy the research needs or that inadequate measures have been used. It may also be the 
case that no previous measure has been proposed for the targeted entrepreneurial con-
struct although evidence from the literature and the field suggests its importance.
A well-expressed theoretical base that clearly specifies the content domain of the con-
struct is crucial for all subsequent steps (Netemeyer et al., 2003). Based on the specified 
domain of the construct the generation of an item pool takes place.
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 55
Step 2: Item pool generation
After specifying the domain and defining the purpose of the construct a researcher gen-
erates a large pool of items that are candidates for eventual inclusion in the scale (De-
Vellis, 2003) and that captured the domain (Churchill, 1979) of the new entrepreneurial 
construct. An initial list of items can be generated applying techniques that are usually 
productive in exploratory research and include a literature review, examination of exist-
ing scales related to the new construct (Churchill, 1979) and expert judges (DeVellis, 
2003; hardesty & Bearden, 2004). Additional items can be created based on interviews 
and focus groups conducted in the first step. A researcher can get ideas for item de-
velopment also from the analysis of interviews, web pages, statements and lectures of 
entrepreneurial expert and practitioners effectuated in the first step as well as from the 
popularly-scientific press.
Writing items is an art (Nunnally & Bernstein, 1994). however, good items can be writ-
ten following some basic rules (e.g. Clark & Watson, 1995; DeVellis, 2003; Netemeyer 
et al., 2003; Nunnally & Bernstein, 1994). Items should be clear, simple and short. A 
researcher should pay attention also to the reading level ease of the target population, 
e.g. entrepreneurs and entrepreneurial students, should avoid multiple negatives, double 
barreled sentences, and use of jargon or trendy expressions. Items should be written in 
a way to ensure variability in responding. The initial item pool should include a large 
number of items since at this stage oveinclusiveness is preferred to underinclusiveness. 
DeVellis (2003) argues that it would not be unusual to begin with a pool of items that is 
three to four times as large as the final scale. 
Items serve as evidence of content validity. The appropriateness of items and their relat-
edness to the specified construct is assessed in the next step in which the evaluation of 
content validity begins.
Step 3: Content validity evaluation
Content validity refers to the degree to which elements of a measurement instrument 
are relevant to and representative of the targeted construct for a particular assessment 
purpose (haynes et al., 1995). Elements of a measurement instrument include items, 
response formats and instructions. These elements represent key parts of a questionnaire 
and questionnaires are a widely used data collecting format in entrepreneurship research 
(Chandler & Lyon, 2001; Crook et al., 2010). Based on its importance, the development 
of the questionnaire is discussed in the next step. here we expand on content validity 
evaluation. 
The representativeness criterion has two meanings. First, it refers to the degree to which 
the domain has been sampled, i.e. if the elements of the construct are proportional to 
the facets of the target construct (haynes et al., 1995; Loevinger, 1957; Netemeyer et 
al., 2003). Accordingly, Nunnally & Bernstein (1994) argue that item pools should be 
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201256
regarded as samples of content and they should be evaluated in terms of how well they 
sample the implied domain and how relevant are they for the target construct (DeVel-
lis, 2003). Thus, all items should be reviewed by judges (hardesty & Bearden, 2004), i.e. 
knowledgeable people in the content area (DeVellis, 2003) to check for their content 
validity. Reviewers of items should be chosen broadly (Nunnally & Bernstein, 1994). In 
an entrepreneurship context we should include end users (scholars that would admin-
ister items of the new construct), subject-matter experts (e.g. entrepreneurship profes-
sors, business angels, venture capitalists, representatives of technology parks, business 
incubators, chambers of trade, and financial institutions), psychometricians, and rep-
resentatives of those who will be taking the test (e.g. entrepreneurs, entrepreneurship 
students, nascent entrepreneurs). At this stage suggestions for modification, addition, 
and exclusion of items are often made. The evaluation of items should be in a formal 
note. Zaichkowsky (1985) proposed an evaluation method, whereby each item is rated 
as “clearly representative”, “somewhat representative” or “not representative” of the new 
construct.
The second meaning of the representativeness criterion refers to the degree to which 
elements are representative of the construct (Netemeyer et al., 2003). Thus, it is recom-
mended to take notes of all the comments that judges express in regard to representative-
ness, clarity and wording of items, clarity of instruction, response formats, and sequence, 
length, and appearance of the assessment instrument. Based on these comments and the 
evaluation of the representativeness of items, the measurement instrument should be 
revised. Despite experts̀  opinions on retention, modification or exclusion of items, De-
Vellis (2003) warns that the final decision to accept or reject the advice of experts is the 
responsibility of the scale developer.
In order to increase the legitimacy of the new construct, information regarding the rep-
resentativeness, relevance, and evaluation of the measurement instrument should be re-
ported in papers. Besides for the publication reason, tracking all the stages (generation, 
modification, addition, and elimination of items) is useful also for researchers them-
selves. It should be noted that the evaluation stage should be repeated as long as items 
and other elements of the assessment instrument are seen as the most appropriate. This 
may involve several periods of item writing, followed in each case by their analysis (Clark 
& Watson, 1995). 
These first three steps are crucial in scale development since they influence construct va-
lidity. A researcher might not be able to assess construct validity if a construct is not pre-
cisely or appropriately defined or if measurement items are not sufficiently representative 
or relevant of the new construct. C. R. Reynolds (2010) warns that it is not appropriate 
to simply write a set of items and begin research on a construct since this omits key first 
steps, namely conducting the necessary research on the measurement instrument itself, 
modifying it appropriately, trying out the items again – and again if necessary – and 
then applying the instrument to the research on the construct in question and making it 
available to other researchers. 
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 57
Phase 2: Representativeness and appropriateness of data collection
The second phase regards the representativeness and appropriateness of data collection 
and is comprised by the four steps that follow the previous three steps: (4) questionnaire 
development and evaluation, (5) translation and back-translation of the questionnaire, 
(6) pilot study performance, and (7) sampling and data collection. Figure 4 represents the 
scale development procedure with the second phase marked in different color.
Figure 4: Phase 2 – Representativeness and appropriateness of data collection
Step 4: Questionnaire development and evaluation
A common method for data collection in entrepreneurship research is via questionnaires 
(Chandler & Lyon, 2001; Crook et al., 2010). Besides the commonly used paper-and-pencil 
version, the usage of web-based surveys is increasing. A properly designed questionnaire 
facilitates gathering relevant data for the study. Dillman et al. (2009) proposed the tailored 
design method for internet, mail, and mixed-mode surveys. Following rules for establish-
ing trust, increasing the benefits of participation, and decreasing costs of participation a 
researcher gets more reliable data and improves response rates. Examples of these rules 
are: ensure confidentiality, provide information about the survey, ask for help, say thank 
you, make it convenient to respond, and send remainders (Dillman et al., 2009).
Pre-testing is generally agreed to be an indispensable stage in survey development (Col-
lins, 2003; Dillman et al., 2009; Presser & Blair, 1994; Presser et al., 2004), thus the survey 
should be pre-tested with people who have specialized knowledge on some aspects of 
questionnaire quality (Dillman et al., 2009). In the entrepreneurship research setting, we 
suggest also to administer the pilot version to the target population. This can be carried 
out as a “think aloud” protocol where the researcher takes notes of comments raised by 
the evaluator.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201258
When developing the measurement instrument that better suits the targeted construct, 
e.g. a questionnaire, we suggest including scales of validated related constructs for an 
initial comparison among constructs already in the pilot testing stage, that is presented 
in Step 6. Besides validated scales of related constructs, we suggest also the inclusion of 
a social desirability scale. This serves to detect potential distorted item responses due 
to the inclination of individuals to represent themselves in a way that society regards as 
positive (DeVellis, 2003). The most widely used social desirability scale is the Marlowe-
Crowne social desirability scale (Crowne & Marlowe, 1960). Since it contains 33 items, 
several shorter versions have been developed, e.g. a 10-item measure by Strahan and 
gerbasi (1972) and a 13-item measure by W. M. Reynolds (1982). With the inclusion of 
a social desirability scale the Differential Reliability Index can be calculated in order to 
maximize the content saturation of the item in relation to its saturation with a desirabil-
ity scale (jackson, 1971).
Step 5: Translation and back-translation of the questionnaire
The cross-cultural equivalence is a prerequisite for comparison across cultural and eth-
nic boundaries (hui & Triandis, 1985). Brislin (1970) suggested several methods aimed 
at putting the same test in different languages while preserving the same ideas across 
the linguistic boundaries. One of these methods is back-translation by which a meas-
urement instrument should be translated and back-translated in the original language. 
The equivalency of the original version and back-translated version should be evalu-
ated.
Step 6: Pilot study performance
At this stage the first data collection and evaluation begins. Before engaging in the large 
data collection for the assessment of psychometric properties of the new construct, it is 
important to conduct a preliminary study on a moderately sized sample on the targeted 
population, e.g. on 30-50 entrepreneurs. Although students are a convenient sample for 
preliminary research (Netemeyer et al., 2003), a researcher cannot always draw reliable 
conclusions from their results. The reason is that the construct may have different prop-
erties in different samples (Clark & Watson, 1995). C. R. Reynolds (2010) also argues that 
collegiate samples, especially from a single university or campus, are rarely representa-
tive of anything beyond students on that campus. Thus, we suggest conducting the pilot 
study on the target population, i.e. entrepreneurs, entrepreneurial ventures and others. 
Entrepreneurship students would be a relevant sample if a research tries to evaluate en-
trepreneurial intentions or other pre-venturing factors.
This mini-study has two purposes. On the one hand, with a pilot study on the target 
population a researcher tests the proposed questionnaire in order to identify potential 
problems with the questionnaire (Dillman et al., 2009). On the other hand, a researcher 
obtains a pre-result of the reliability of the new measure (Netemeyer et al., 2003) and 
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 59
examines the boundaries of the target construct and correlations of the target construct 
with existing measures (Clark & Watson, 1995). In doing so, a researcher is able to detect 
flaws or problems with the measure and hopefully demonstrate that the new construct 
is distinct from other related constructs. Clark and Watson (1995) note that all too often 
researchers discover late in the process that their new measure correlates too highly with 
an existing measure.
Step 7: Sampling and data collection
With a sampling process we aim at obtaining a representative portion of some whole 
to afford valid inferences and generalizations to it (Pedhazur & Pedhazur Schmelkin, 
1991). Thus, it is important to carefully undergo the sampling process. At this stage the 
researcher aims at collecting data for the evaluation of the measurè s factor structure and 
for the subsequent analysis of various types of validity (hinkin, 1998). Random sam-
pling is the most common type of sampling used in self-administered surveys (Dillman 
et al., 2009). In random sampling every member of the sample frame has an equal chance 
of being selected.
At this stage it is crucial to administer the questionnaire to the actual population 
of interest (hinkin, 1998). Thus, the researcher should choose respondents from a 
database of entrepreneurs or entrepreneurial ventures. hinkin (1998) also proposes 
that when developing a scale, it is necessary to use several independent samples. 
Besides sampling in one country, cross-cultural samples should be taken into ac-
count.
While an adequate representation of the target population in the sample is crucial 
for inferences to the whole target population, also the size of the sample should be 
considered. There is no established rule about the size of the sample. Still, two main 
recommendations should be followed. First, the sample of subjects should be large 
(DeVellis, 2003) and second, as the number of items increases, the number of respond-
ents should increase too (hinkin, 1998). Some recommendations for item-to-response 
ratios suggest a 1:10 ratio (Schwab, 1980), by which 100 respondents are needed for 
a 10-item measure. Similarly, DeVellis (2003) proposes that if only a single scale is 
to be extracted from a pool of about 20 items, fewer than 300 subjects might suffice. 
hinkin (1998) suggests that for factor analysis a sample of at least 200 respondents is 
required.
Phase 3: Statistical analysis and statistical evidence of the construct
In the final phase statistical analysis and statistical evidence of the construct is per-
formed. This is achieved by the last three steps: (8) dimensionality assessment, (9) reli-
ability assessment, and (10) construct validity assessment. Figure 5 represents the scale 
development procedure with the third phase marked in different color.
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201260
Figure 5: Phase 3 – Statistical analysis and statistical evidence of the construct
Step 8: Dimensionality assessment
Dimensionality of a measure is concerned with the homogeneity of items and is defined 
as the number of common factors needed to account for the correlation among items 
(Netemeyer et al., 2003). homogeneity indicates whether the scale items assess a single 
underlying factor or construct (Clark & Watson, 1995). A measure that is considered uni-
dimensional has a single facet or dimension, which means that its items underlie a single 
factor. Accordingly, a multidimensional measure has multiple facets or dimensions and 
its items tap more than one factor (Netemeyer et al., 2003). 
Dimensionality can be assessed with either exploratory factor analysis (EFA) or con-
firmatory factor analysis (CFA) or both. EFA is commonly conducted in the initial stage 
of scale development. When conducting EFA it is assumed that a researcher has a limited 
idea of the new measurè s dimensionality (Netemeyer et al., 2003). In CFA the number 
of factors, the factor structure and the relationship among factors are set a priori. CFA 
evaluates whether the hypothesized factor model fits the data or not (Netemeyer et al., 
2003). Unidimensionality of a scale can be assessed by average interitem correlations 
taking in consideration also the range and distributions of correlations. Interitem cor-
relations should fall in the range of 0.15 and 0.50 and should cluster narrowly around the 
mean value (Clark & Watson, 1995).
Step 9: Reliability assessment
After the dimensionality of the new measure has been set, the researcher proceeds with 
assessing measurè s reliability. Scale reliability is the proportion of variance attributable 
to the true score of the latent variable (DeVellis, 2003). Thus, it deals with that portion of 
A. SLAVEC, M. DRNOVŠEk  |  A PERSPECTIVE ON SCALE DEVELOPMENT IN ENTREPRENEURShIP ... 61
measurement that is due to permanent effects that persist from sample to sample (Nete-
meyer et al., 2003). Scale reliability can be assessed with several methods, e.g. temporal 
stability, split-half reliability and internal consistency. The temporal stability or test-retest 
reliability refers to correlations between the same persoǹ s score to the same test at two 
different points in time (DeVellis, 2003). In split-half reliability the researcher divides the 
set of items into two subsets and correlates the subsets to assess the reliability (DeVellis, 
2003). But the single most widely used method for item selection in scale development 
is some form of internal consistency analysis (Clark & Watson, 1995). Internal consist-
ency refers to the homogeneity of items within a scale (DeVellis, 2003). Thus, a scale is 
internally consistent if items of a scale are highly intercorrelated. high inter-item cor-
relations suggest that items are measuring the same construct (DeVellis, 2003). The most 
commonly employed internal consistency measure is the Cronbach s̀ (1951) coefficient 
alpha. Cronbach alpha represents the mean of all split-half coefficients resulting from 
different splittings of a test (Cronbach, 1951). Different suggestion have been proposed 
for the acceptable level of coefficient alpha, but an alpha above 0.8 constitutes a reliable 
measure (Carmines & Zeller, 1979; Clark & Watson, 1995). Besides this, hair et al. (2010) 
suggest that the generally agreed upon lower limit for Cronbach’s alpha is 0.70, although 
it may decrease to 0.60 in exploratory research. Moreover, internal consistency can be 
evaluated also with item-to-total correlations and inter-item correlations. The item-to-
total correlation is the correlation of the item to the summated scale score and items 
with correlations below 0.50 with the scale score should be eliminated (hair et al., 2010). 
The inter-item correlation refers to correlations among items and items with correlations 
below 0.30 with other items should be excluded (Flynn, Schroeder, & Sakakibara, 1994; 
hair et al., 2010). When evaluating items, Clark and Watson (1995) propose that items 
that have highly skewed and unbalanced distributions should be removed, while it is 
reasonably to retain items that show a broad range of distributions.
If a social desirability scale is included in the study of the new construct, the Differential 
Reliability Index (jackson, 1971) can be evaluated. This serves to identify answers to 
items that are biased by social desirability. It may be that some respondents try to present 
themselves in a more social desirable way. For each item, the correlation between the 
scale item and the total desirability score is subtracted from the correlation between that 
scale item and the total scale score. The square root of the remainder is the proportion of 
true content variance for any given item (Leavitt & Walton, 1975).
Step 10: Construct validity assessment
In the final step of scale development, validity of the measure is to be assessed. Validity 
has been given several meanings and consensus has not been reached for the categori-
zation of all the types. Construct validity seems the most important one and for some 
scholars construct validity subsumes all categories of validity (haynes et al., 1995; Ne-
temeyer et al., 2003). Construct validity refers to the extent to which any measuring in-
strument measures what it is intended to measure (Carmines & Zeller, 1979; Netemeyer 
et al., 2003) in the context in which it is to be applied (Nunnally & Bernstein, 1994). Put 
ECONOMIC AND BUSINESS REVIEW  |  VOL. 14  |  No.  1  |  201262
in other words, a researcher measures the extent to which a set of measured items actu-
ally reflects the theoretical latent construct those items are designed to measure (hair et 
al., 2010). Among other types, construct validity is investigated through convergent and 
discriminant validity. 
Convergent validity is the extent to which responses from alternative measurements of 
the same construct share variance (Schwab, 1980). That is, a measure has convergent 
validity if independent measures of the same construct converge, or are highly corre-
lated (Netemeyer et al., 2003). Discriminant validity refers to the degree to which two 
measures designed to measure similar, but conceptually different constructs are related 
(Netemeyer et al., 2003).
Convergent and discriminant validity can be assessed with different methods, e.g. by 
multitrait-multimethod matrix (MMTM matrix) (Campbell & Fiske, 1959; Netemeyer 
et al., 2003) and by factor analysis. While MMTM is useful in assessing convergent and 
discriminant validity, it has been criticized by several researchers because of unrealis-
tic assumptions (e.g. Bagozzi, Yi, & Phillips, 1991). Factor analysis, specifically CFA, is 
strongly recommended at this stage. In CFA a researcher should examine the loadings 
of items on factors. Items that load on a factor relatively strongly and weakly on other 
factors are candidates for retention (Clark & Watson, 1995). Factor loadings should be 
statistically significant and standardized loadings should be above 0.50 (ideally 0.70) 
(hair et al., 2010). Additionally, average variance extracted (AVE) can be analyzed. AVE 
is a summary indicator of convergence. An AVE of 0.50 or higher suggests adequate 
convergence. On the other hand, discriminant validity refers to the absence of correla-
tion between measures of unrelated constructs (DeVellis, 2003), therefore the presence 
of cross-loadings would present a discriminant validity problem (hair et al., 2010). Items 
that loaded on different factors should be eliminated.