Scale Construction Proposal – Adventure vs. Safety Personal Construct

1. Introduction

In data collection, a scale refers to the closed-ended questionnaire used if the items are
clustered according to their similarities and internal consistency (Hogan, 2019). Constructing a
scale requires considerations such as the homogeneity of the scale, the ability of the respondents
to understand what’s being measured, plus its accuracy in measuring the intended constructs,
among others (Hogan, 2019). Items are clustered under one scale if they measure one construct
or a single concept of the construct (Groth-Marnat, 2009). In some cases, the instrument may be
developed to measure a simple construct; hence the questions may not be grouped. However,
sometimes more than one concept is measured in one instrument, in which case the related items
are grouped to form scales (Groth-Marnat, 2009).
Scales are constructed based on existing literature, either by improving existing scales or
from theories (Hogan, 2019). The measured constructs and concepts are already defined if the
scale is developed to improve existing data collection instruments. Therefore, the development
involves revising or adding questions to ensure the construct is accurately and exhaustively
measured (Groth-Marnat, 2019; Hogan, 2019). If the instrument is being developed from the
beginning, the constructs are derived from existing theories, in which the concepts are studied to
facilitate the creation of the items. Under each construct or concept being measured, the
developer comes up with statements that exhaustively and accurately represent the main
summary points. The items are clustered based on how they relate to one another and their
coverage of the examined concepts (Hogan, 2019). The final item list is determined after various
psychometric tests, which ensure that the scale will measure what it created to measure (Sürücü
& Maslakçi, 2020).

4

The psychometric tests also measure the ability of the responses to reflect the
respondent’s intended response (Sürücü & Maslakçi, 2020). Responses to a scale are developed
to reflect the closest representation of the respondent’s intended response. Given that scales are
closed-ended questions, the responses are limited to choices, in most cases five; therefore, they
should be created to facilitate the accurate representation of the intended response (Hogan,
2019). The five response choices, if ordered, are referred to as a Five-point Likert scale, in which
the respondents indicate the level to which they agree with the presented statements in five
points ranging from ‘strongly disagree’ to ‘strongly agree’ (Hogan, 2019).
This paper examines one of psychology’s most important personal constructs: the
adventure vs. safety construct, including conceptual definitions and related concepts. It then
reviews the literature and existing tools that measure the construct, provides justification for the
selection of the scale, and proposes how the tool will be constructed. It will also provide twenty
sample items representing the construct, a method of running an item analysis procedure using
the scale, and present modalities of establishing the scale’s validity, including construct and
criterion-related validity.
1.1. Adventure vs. Safety Personal Construct
The construct to be measured in the constructed instrument is ‘adventure versus safety,’
which is a personal construct dimension. The personal construct psychology was introduced by
psychologist George Kelly (1905–1967) in a two-volume book, The psychology of personal
constructs. The book is divided into a personality theory for its first volume and clinical
diagnosis and psychotherapy for the second. Kelly (2001) introduced the theory of personal
construct psychology, in which he proposed that people may create personal constructs to
organize their ongoing experiences and prepare for the anticipated resultant events. Kelly (2001)

5
further posits that people develop unique personal constructs to organize experiences coherently
and in ways easing understanding and preparing for the resultant anticipated events.
In personal construct psychology, individuals develop unique constructs for different
experiences, which means that one person can develop an infinite number of constructs (Kelly,
2001; Kelly, 2017). Additionally, the theory posits that although only one true reality exists,
people perceive this reality from their perspectives; hence, different people will develop varying
constructs from a similar experience. Kelly (2017) classified the differences in construct
formation as constructive alternativism.
Additionally, Kelly (2001) theorized that personal constructs are bipolar, indicating that
they must exist in opposites. The ‘adventure vs. safety’ construct is an example of the bipolar
nature of personal constructs. In personal construct psychology, a person’s choice is based on a
dichotomous construct. Based on this choice, the individual can expect a possible extension or
definition of an alternative portion of the construct as part of their personality. For example, the
‘adventure vs. safety’ construct is dichotomous and is defined by the individual’s perception of
their choice of experiences or behavior. By choosing the standard option, the person’s choice is
viewed as safe and mundane, which is transferred to their personality. Conversely, the person
eager for exciting and daring experiences is viewed as adventurous, which translates to their
personality when in that particular domain (Kelly, 2001). The domain is always an important
consideration when assessing the different personalities of individuals. If, for example, the
domain in question is traveling, a safe traveler may prefer to plan before leaving and commit to
the plan when choosing activities. In contrast, an adventurous traveler may only plan after some
time and will choose random activities during their travel.

6
Applying the ‘safety vs. adventure’ personal construct transcends different dimensions. In
every dimension where individuals have to choose between the safe or adventurous option, their
personality is presented by these choices. The theory of personal construct helps us understand
how different people view their environment, including the choices they have to make. The
perception we get from observing and understanding the person’s choice about themselves helps
us define their personality in the presented bipolar construct (Kreber et al., 2003). Therefore, the
perception of the two opposites of the construct differs for individuals based on context and
personal experiences. The views and beliefs held by different people form the basis of forming
their personalities, which they then use to define their experiences and observations (Kelly,
2017).

2. Literature Review

For a long time, psychological tests have been labeled as either objective or projective.
Bornstein (2007) affirms that the labels are no longer relevant and proposes labeling based on the
psychological processes resulting in the individual’s response to the test stimuli. In this case,
attribution, self, or stimuli-attribution determine how people respond to the test stimuli.
Therefore, the tests should be either self-attributed or stimulus-attributed (Bornstein, 2007). The
personal construct theory introduced the notions of personalities and cognition in personality
psychology, forming the basis for modern psychotherapy (Kelly, 2017). In the original theory,
Kelly (2001) posited that psychological evaluations could be conducted based on personal
constructs, which are the bipolar schemas in which people interpret the world (Kelly, 2017).
The personal construct theory formed the basis of cognitive therapy and the repertory
grid interviewing, which facilitates the patient’s recognition and analysis of their “own”
constructs (Kerber et al., 2003). In cognitive therapy, the psychotherapist avoids interfering with

7
or interpreting the patient’s constructs, facilitating the patient to find their unique construct
(Kelly, 2001; Kelly, 2017). With the success of cognitive therapy in psychotherapy, the repertory
grid interview technique has been recognized as an effective strategy for examining the personal
constructs of individuals. The repertory interview can be presented in different formats,
including card slots, oral administration, or a grid or matrix (Kerber et al., 2003).
In the repertory grid, the matrix rows represent the constructs of the interviewee, while
the columns and cells are the elements and their numeric positioning within the construct. The
software can analyze the patient’s construct by producing reports that include charts. Given the
infinite number of elements within a given construct, the patient is first asked to choose the
elements related to the nature being uncovered through therapy, three of the elements are then
randomly selected, and the patient is asked their position; the procedure is repeated severally
until the patient exhausts their constructs including “those they were unaware of their existence”
(Kelly, 2017). The repertory grid has been adapted for use in numerous fields and has shown
promising results in identifying individuals’ positions toward the examined constructs.
In one study that applied the repertory grid, Kreber et al. (2003) applied the theory of
personal construct and the repertory grid to undergraduate students’ learning evaluation. The
authors adapted the repertory grid based on the assumptions made in the personal construct
theory repertory grid and based them on undergraduate learning experiences. The adapted
repertory grid was then used as a research tool by university educators as a classroom assessment
technique, and an evaluation of its application and sustainability was conducted. The evaluation
established that the adapted version is a useful representative of classroom assessment and can be
used to map science learning and, to some extent, teaching. The version of the repertory grid
adapted by the authors is called the role construct repertory test (RRT). The authors concluded

8
that it was more beneficial to use the repertory grid in classroom assessment than standard
assessments because of the result’s ability to reflect and map the concepts requiring more or less
attention (Kerber et al., 2003).
In a different study, Walker et al. (2020) used the repertory grid to examine participants’
perceptions of psycho-education for psychosis. The assessment using the grid was conducted
before and after the participants received a group intervention regarding forensic services. The
20 participants were asked to discuss their views regarding the presumed changes in people with
psychosis following attending a psychoeducation group intervention. During each of the two
assessments, the participants were assessed using structured interviews constructed based using
the repertory grid. The authors used the Gridsuite program to analyze the data obtained from the
assessments. The results indicated a change in the participant’s experiences of the group
interventions caused by more than one factor. The authors concluded that the repertory grid
provided an alternative strategy for establishing the effect of psychological interventions on
mentally disordered offenders.
Horley (2012) also applied the personal construct theory to define and map human
values. The author developed a framework for categorizing human values and beliefs as either
primary or peripheral constructs. The author utilized the lack of a theoretical framework for
framing human values as a motivator to use the personal construct theory to map human values.
The theory of personal construct theory had been previously suggested as a possible theoretical
foundation for human values. The author utilized these existing arguments and found it possible
to map human values against the personal construct theory (Horely, 2012).
In his early work, Horley (2009) explored the application of the personal construct theory
in social and forensic psychology. In this particular work, the author utilized the personal

9
construct theory to examine its possible application in examining the behavior of sex offenders
and their reaction to tailored therapy based on identifying their “personal constructs.” The author
established that utilizing the theory links the relationship between how different people perceive
their world and sex offenders’ behavior. Understanding their personal experiences and how they
affect their view of their world and the resultant behavior is possible for sex offenders. It can be
pursued to tailor-make psychotherapy approaches targeted to the population. Based on Kelly’s
(2001) work on cognitive therapy, sex offenders can learn to perceive their experiences in their
personal constructs, which could, in turn, affect their choices and behavior. Understanding one’s
self is key to expecting behavior change. Using the theory, the author established that sex
offenders might be able to change their views of the world and change their behavior with
cognitive therapy (Horley, 2009).

3. Existing Tools

The ‘adventure vs. safety’ construct is just one of the infinite bipolar personal constructs.
As a result, no existing tools were developed to measure the ‘adventure vs. safety’ construct.
Although different tests exist for measuring adventure and safety or risk separately, they are not
related to the personal construct theory, which was the proposal’s focus. Some tests measure
personal constructs and are closely related to the ‘adventure vs. safety’ construct. They include
the depressive experiences questionnaire (DEQ), the role construct repertory test (RRT), and the
sensation-seeking scale (SSS) and its variants.
The RRT is an adaptation of the repertory grid, also developed by Kelly, alongside the
personal construct theory (Kreber et al., 2003). The RRT is a measure of the application of the
personal construct theory and uses a grid that measures how different people perceive and
interact with the world. The test uses the bipolar nature of constructs as suggested by the

10
personal construct theory. Yet, the measurements concern the interviewees’ perceptions of people
or objects rather than examining their own personalities. The test provides a guided approach for
assessing one’s personality in relation to daily interactions through unscripted descriptions of
one’s opinions. The responses are naturally bipolar, and the interviewees respond to the questions
using the extreme opposite of each construct. For example, interviewees choose between happy
and sad when asked how a certain experience makes them feel (Kreber et al., 2003).
The DEQ measures the interviewees’ behavior and personality in relation to depression
symptoms (Falgares et al., 2018). The DEQ was created in 1976 as a self-report test comparing
the respondent’s self-criticism versus their dependency, both of which are personality traits
associated with depression symptoms. The DEQ does not necessarily measure bipolar constructs,
as suggested in the personal construct theory; however, self-criticism and dependency are
dimensions of a depression-prone personality (Falgares et al., 2018).
The SSS has various variants developed to cover the different limitations arising over
time. The original SSS was created as a risk assessment measure regarding people’s interactions
with the natural environment (Prochniak, 2017). Based on the scale, people respond differently
to the natural environment depending on their sensation-seeking personality trait. Individuals
who prefer novel, thrilling experiences and are willing to incur the risks involved are at the high
end of the sensation-seeking personality, while those who are their opposites are on the bipolar
side of sensation-seeking. The original scale measured four sensation-seeking concepts: thrill and
adventure, experience, disinhibition, and susceptibility to boredom. Because of being a long a
shorter version of the scale covering all four factors, the brief sensation-seeking scale (BSSS)
was developed. The most recent variation of the SSS is the adventure behavior-seeking scale

11
(ABSS), which is also brief, with eight items. It measures the adventure-seeking behavior of
people in the natural environment (Prochniak, 2017).
The ABSS differs from the SSS and BSSS on the response scales. The first two versions
of the SSS utilized bipolar responses, while the ABSS used a five-point Likert scale to measure
the respondents’ adventure-seeking personality. Additionally, the SSS and BSSS measure
individuals’ sensation-seeking and risky behavior linked to thrill, experiences, and boredom,
among others. On the other hand, the ABSS measures adventure-seeking based on environmental
factors, such as water, weather, and gravity threats (Prochniak, 2017).

4. Rationale

The literature review revealed the different applications of the personal construct theory
in education and psychology. The ‘adventure vs. safety’ construct can be useful in different
dimensions, such as tourism and organizational development. However, there needs to be more
literature supporting their application in these dimensions. As a result, there is a need for more
literature and more tests on applying the ‘adventure vs. safety’ construct in these areas. In the
tourism sector, developers and businesses use tourists’ perceived adventure or safety personality
to design marketing plans (). However, there are limited resources for testing both the adventure
and safety proneness of individual guests and potential customers that may help tailor the
different experiences sold to tourists.
Developing an adventure vs. safety construct scale linked to people’s interaction with
nature will be helpful for tourist attraction sites. The scale can be utilized to interview guests and
other potential customers to determine what activities they may be attracted to. Tourism sites can
leverage this information to tailor-make programs for individuals based on their personalities.

12
Additionally, the scales may be useful for guiding the development of marketing campaigns
tailored to people with different personalities.

5. Scale Construction

Constructing the scale will begin with reviewing the theory of personal construct as
presented by Kelly (2001) and any improvements and criticisms of the original work. Reviewing
the theory facilitates understanding the personal construct theory and the fact that people form
constructs based on various factors, such as the environment. Additionally, a specific area of
focus is selected because the perceptions of adventure or safety may change depending on the
circumstances. Some specific areas of study include participation in outdoor activities, career
opportunities, and spending free time, among others. The proposed field of focus will consider
the ‘safety vs. adventure’ construct based on individuals’ choices when participating in outdoor
activities.
The construction of the items will be conducted in two stages. In the first stage, the items
will be categorized based on their relation to adventure and safety (See Table 1). Given the focus
will be on outdoor activities, the questions will be focused on activities that signify safe or
adventurous choices. Items can be static, valent, based on frequency, or valent and frequency-
based. Static items provide descriptions of one’s activities or behavior. Items using a frequency
approach describe behavioral and mental processes, while valence-based items provide reports of
the degree of one’s feelings towards something. Items measuring one’s experiences are static;
those that measure how often something occurs are frequency-based, while those that measure
the respondent’s feelings towards something are valent.
Table 1
Stage 1 Item Construction

13

Adventure Safety
i. I don’t believe in early planning when
headed outdoors

i. I prefer planning for outdoor
activities in advance

ii. I choose my outdoor activities at
random

ii. I will only stick to my previous plan
during outdoor activities

iii. I often experiment with new trails
when hiking

iii. I would never take a new trail when
hiking.

iv. I prefer spontaneous random hikes. iv. I don’t like random events.
v. I will swim if I find a water source
during my hikes

v. I would never risk getting into
random water bodies.

vi. I don’t believe in appropriate gear vi. There is appropriate gear for every

event.

The constructed items will be reviewed in the second state of item development to
determine their appropriateness and language structure. After reviewing the items in Table 1, it is
established that they cover some of the ‘safe vs. adventurous’ practices when outdoors, such as
pre-planning, sticking to the plan, avoiding random forest trails and water bodies, and using the
appropriate gear to avoid injuries. There is a visible repetition of items for both the adventure
and safety sections. The sections can be merged to eliminate these redundancies, given that
individuals’ perceptions are bipolar. Therefore, the perceptions of the respondents can either be
adventure or safety. Combining the two sections resulted in Table 2.
Table 2
Combined Sections in Stage Two of Scale Construction
Planning.

1. I will only participate in previously planned activities
2. I will stick to the previously planned schedule when
   outdoors

14

3. I will only take the planned familiar route when hiking.

Water risk

4. I would never get into unknown water bodies.
5. I would never use wild water bodies even if they seem
   clean
6. I cannot swim in wild water bodies
   Gear/Protectiveness
7. I will only hike if I have the appropriate hiking gear
8. I will only hike if I have the appropriate sun protection

Leaving the items as in Table 2 facilitates the identification of the respondent’s
inclination based on the sums or average score against a set target. For example, for the eight
questions, the maximum score (for a five-point Likert scale, 1 to 5) would be eight and the
maximum 40. Assigning the maximum and minimum scores to the safety or adventure
inclination would facilitate the respondent’s inclination.
The response scales used for measuring a construct are nominal, ordinal, and interval. In
nominal scales, distinct response classifications are assigned numbers as labels. The numbers
used to label the categories cannot be computed and do not hold any significance to the
categories they represent (Fishman & Galguera, 2003). Example: What is your gender? 1. Male.

2. Female. Ordinal scales are categorical responses where the number assigned to each category
   signifies their ranking. An ordinal scale assigns numbers to responses based on their ranking in a
   linear rating scale. Ordinal scales are more informative and allow more measurement options

15
than nominal scales (Fishman & Galguera, 2003). However, the distance between the scales is
unknown. An example of a five-point Likert ordinal scale is a question assessing the level of
satisfaction in individuals with the following responses: Very unsatisfied. 2. Unsatisfied. 3.
Neutral. 4. Satisfied. 5. Very satisfied.
Ordered numerical values with equal and meaningful distances represent interval scales.
An interval scale measures variables within a common scale at equal intervals. However, the
presence of zero in interval scales is arbitrary. The difference between interval and ordinal scales
is the uniformity and meaningfulness of the distance between the assigned values. An interval
scale with a natural zero is referred to as a ratio scale. Based on the questions, the response scale
will be ordinal. The responses will be categorized and ranked from 1 to 5 using a five-point
Likert scale of (1) Strongly Disagree, (2) Disagree, (3) Neither disagree nor disagree, (4) Agree,
and (5) Strongly Agree.

6. Item Analysis

The items will be analyzed by scoring the responses. The responses will be five-point
Likert scales, each representing a unique number between 1 and 5. The average score will be
calculated by summing the item scores and dividing the sum by the number of items. An average
score is preferred because the resultant value is between 1 and 5, which gives a clear picture of
the respondent’s personality between the two extremes.

7. Psychometric Properties

The scale’s internal consistency will be measured using the Cronbach alpha coefficient
and the split-halves method. In this case, Cronbach’s alpha coefficient will be used. It is preferred
for Likert scales and is calculated using a computer program. The instrument will be considered
reliable if the value is 0.7 or higher. Cronbach’s alpha provides a value used to represent the

16
instrument’s internal consistency, which is interpreted as measuring the relationship between the
instrument items. Therefore, the obtained α value will be the instrument’s internal consistency
coefficient.
Measuring the instrument’s validity tests its ability to accurately and exhaustively
measure what it was intended to measure. The instrument’s validity will be measured for the
construct and content. Measuring construct validity involves assessing the instrument’s
representation of the tested concept. The items or questions measuring specific concepts or
constructs should be able to measure the concept accurately. Construct validity can be measured
using factor analysis, which is the most commonly used and is considered a strong measure of
the instrument’s validity. There are two types of factor analysis, confirmatory and exploratory,
applied based on whether the test is newly or previously developed. The instrument developed is
new, meaning exploratory factor analysis will be used. Exploratory factor analysis (EFA) is used
when the factors in the dataset are unknown or not well understood, like in the case of a new test
such as the current one. EFA is the most effective when the items are related, which is the case
for the test.
EFA can be calculated and interpreted using both statistical output and graphs. In this case,
the analysis software will calculate a scree plot. A scree plot (See Figure 1) demonstrates the
number of factors (constructs) that should be maintained in a scale. A scree plot constitutes an
elbow-shaped curve, plotted using eigenvalues and the number of factors on the y-axis and x-
axis, respectively. The number of factors is determined by observing the slope of the curve to
determine the leveling-off point (the ‘elbow’ marked with a red line)
Figure 1
An example of a Scree plot

17

Content validity measures the instrument’s coverage of all the relevant concepts of the
intended construct (Sürücü & Maslakçi, 2020). In constructs that cannot be measured directly,
content validity measures if the items have exhausted all the elements of the measured construct.
To measure content validity, several experts in the field of the construct being measured may
review the instrument items to determine how well it is being assessed. The panel of experts can
be selected based on their knowledge or experience of the construct and given questionnaires
assessing the perceived relevance of each item. The instrument’s content validity ratio (CVR)
and content validity index (CVI) is based on the reviews and responses to the provided
questionnaire. CVR is calculated for each item as the ratio of panelists considering the item
relevant. The average CVR provides the CVI. CVR is calculated using the formula: CVR = , N is
the total number of panelists; Ne is the total number of panelists identifying the item as essential.
The closer to 1 the CVR is, the more correct the item is. However, a CVR> 0.78 is required for
the item to be valid. If the CVR is less than 0.7, the item is discarded. While a CVR value
between 0.7 and 0.79 indicates a need for revising the item. The average CVR produces the CVI.

18

8. Conclusion

[Insert here]

19

9. References

Blanton, H., & Jaccard, J. (2006). Arbitrary metrics in psychology. The American
psychologist, 61(1), 27–41. https://doi.org/10.1037/0003-066X.61.1.27
Bornstein, R. F. (2007). Toward a process-based framework for classifying personality tests:
Comment on. Journal of Personality Assessment, 89(2), 202-207.
Falgares, G., De Santis, S., Gullo, S., Kopala-Sibley, D. C., Scrima, F., & Livi, S. (2018).
Psychometric aspects of the Depressive Experiences Questionnaire: implications for
clinical assessment and research. Journal of Personality Assessment, 100(2), 207-218.
https://doi.org/10.1080/00223891.2017.1282493
Flake, J. K., Pek, J., & Hehman, E. (2017). Construct validation in social and personality
research: Current practice and recommendations. Social Psychological and Personality
Science, 8(4), 370-378. https://doi.org/10.1080/00223890701518776
Groth-Marnat, G. (2009). Handbook of psychological assessment. John Wiley & Sons.
Hogan, T. P. (2019). Psychological testing: A practical introduction. John Wiley & Sons.
https://books.google.com/books?hl=en&lr=&id=K7s6EAAAQBAJ&oi=fnd&pg=PA3&dq
Horley, J. (2009). Sexual offenders: Personal construct theory and deviant sexual behavior (1st
ed.). Routledge. https://doi.org/10.4324/9780203929810
Horley, J. (2012). Personal construct theory and human values. Journal of Human Values, 18(2),
161-171. https://doi.org/10.1177/0971685812454484
Kelly, G. A. (2001). The psychology of personal constructs: Theory and personality. Routledge.
https://doi.org/10.4324/9780203405970

20
Kelly, G. A. (2017). A brief introduction to personal construct theory. Costruttivismi, 4, 3-25.
doi:10.23826/2017.01.003.025. https://www.aippc.it/wp-
content/uploads/2019/04/2017.01.003.025.pdf
Kreber, C., Castleden, H., Erfani, N., Lim, J., & Wright, T. (2003). Exploring the usefulness of
Kelly’s personal construct theory in assessing student learning in science
courses. Teaching in Higher Education, 8(3), 431-445.
https://doi.org/10.1080/13562510309394
Meyer, G. J., & Kurtz, J. E. (2006). Advancing personality assessment terminology: Time to
retire” objective” and” projective” as personality test descriptors. Journal of Personality
Assessment, 87(3), 223-225. https://doi.org/10.1207/s15327752jpa8703_01
Próchniak P. (2017). Adventure Behavior Seeking Scale. Behavioral Sciences 7(2), 35.
https://doi.org/10.3390/bs7020035
Reynolds, C. R., Altmann, R. A., & Allen, D. N. (2021). The problem of bias in psychological
assessment. In Mastering modern psychological testing (pp. 573-613). Springer, Cham.
https://doi.org/10.1007/978-3-030-59455-8_15
Sürücü, L., & Maslakçi, A. (2020). Validity and reliability in quantitative research. Business &
Management Studies, 8(3), 2694-2726. https://doi.org/10.15295/bmij.v8i3.1540
Walker, H., Trenoweth, S., Martin, C., & Ramm, M. (2012). Using repertory grids to establish
patients’ views of psycho-education. Journal of Psychology & Psychotherapy, 3(1). 1-8.
doi: 10.4172/2161-0487.1000108