LJMap.com < Validity
and Reliability
Scott Bristol's 'Values@Hand'.
Validity and Reliability
-
Summary: Life Journey Map® values
measurement methodology challenges many of the popular values measurement
techniques. During the ranking process the user rank
orders a randomly generated list of 130 values, presented in sets of four. Value
ranking scores are the sum of two complete rankings. Validity: 71% of the
users report 'High Degree of Fit' upon seeing their top 6 priority values at the
end of the values ranking process. Reliability: The between test 'Mental
Model' -rank score correlation is .91. (n=2500, 46% men, 54% women).
Life Journey Map® Values
Measurement Methodology
Defines these conditions as necessary for Values
Measurement:
- Values Measurement is based on a values ranking
process.
- The individual is asked to rank order three or more
distinct values at a time (forced ranking).
- The individual is presented with a full domain of
distinct values to rank order.
- The full domain of values is presented randomly to
avoid any biasing.
- The affective (feeling) nature of values allows
them to be either means or ends.
- The individual can discard values from the ranking
process.
- The value definitions shown while ranking are the
value definitions in their totality.
- Value scores relate directly to the individual’s
ranking process.
- An immediate feedback loop at the end of data
collection asks the individual to confirm the validity of his or her
values ranking results.
Considers these techniques inappropriate for
Values Measurement:
- Focusing on small set of values. (Content Validity Error)
- Absence of a feedback loop at the end of data
collection that presents initial results and asks the user to evaluate
relevancy of results. (Criterion-related Validity Error)
- Using a Likert scale when measuring values. (Construct Validity
Error)
- Using Polarity scale when measuring values. (Construct Validity
Error)
- Using Forced Choice when measuring values. (Construct Validity
Error)
- Presenting the full domain of values in a fixed
sequence (non-random) for ranking. (Construct Validity
Error)
- Presenting partial value definitions during
individual scoring that then map to different value definitions for
reporting results. (Construct Validity Error)
- Classifying and measuring separately ‘means
values’ and ‘goal values’. (Construct Validity Error)
Validity
Are we measuring what we think we are measuring?
There are three types of validity: 1) content, 2) criterion-related,
and 3) construct, each attends to a different question (Kerlinger,1973):
1. Content Validity: Do the number of elements
(values) being measured adequately represent the full domain of all
potential elements (all human values)?
- Naturally occurring full linguistic domains appear to have an upper
limit of 500 distinct words (Berlin, 1992).
- According to Life Journey Map's®
definition, there are hundreds of values (Rokeach,
1968).
- Using Christian biblical document analysis, Hall and Tonna were the
first identify a large domain of values (125). (Hall,
1986; Tonna, 1995, 1997)
Life Journey Map® methodology:
- Measures 130 different 'elemental values', each value is defined by
3 different synonyms thus drawing on a full domain of up to 390 value
words.
- Measurements indicate that people operate within a personal domain
of values-a Field of Meaning with an average size=34 different values.
- As a self-correcting document analysis methodology asserts that any
written phrase with a value orientation should equate to three or less
'elemental values'. If not, a new 'elemental value' needs to be added
to the list of measured values.
Common Errors in Values Measurement:
- Most values measurement tools are implicitly biased in that they
fail to present a full domain of values that can be cross checked via
document analysis.
- Brainstorming values fails to generate a full domain of values, it
will only generate the individual's or team's Field of Meaning.
2. Criterion-related Validity: Does the
measurement predict the external variables or criteria? There are two
criterion we are interested in:
a) Does the measurement process result in people
identifying a set of priority values?
Life Journey Map®
methodology:
In response to the question: "According to your own experience and self awareness do these 6
statements reflect the values that you believe and experience as being
most meaningful, significant, and important to you at this time?"
From a sample of n=2500; 46% men, 54% women:
- 71% report a "High Degree of Fit"
- 28% report a "Moderate Degree of Fit"
- 0% report a "Low Degree of Fit"
- 0% report "No degree of Fit"
- 1% report "Cannot Answer"
Common Errors in Values Measurement:
- Non-interactive measurement techniques, such as paper and pencil,
fail to present immediate results to the participant and fail to
collect participant validation of the results.
b) Do the priority ranked values predict 'value
behavior': spontaneous speech and
problem-solving orientation?
Life Journey Map® methodology:
Consultant observations indicate 'yes'. No systematic study has been
completed at this time.
3. Construct Validity: What factors or constructs account for variance in measurement?
Life Journey Map® methodology:
- Uses a ranking schema: important to match mathematical schema to the
phenomena (Lakoff
& Núñez, 2000).
- Uses value definitions in the values ranking process in their
totality, neither adding or removing words from reported results.
- Individual's can choose to discard or remove values from their
ranking domain thus insuring the the 130 domain of values measured is
culturally relevant over a large sample of users.
Common Errors in Values Measurement:
- Using a Likert scale when measuring values.
Example: Rate the degree to which you personally identity with each of
the following ‘value statements’. Use a Likert Scale: 1 = No
Identification, 2 = Low, 3 = Moderate, 4 = High; 5 = Very High
Identification. This process fails to rank order different values,
consequently it is an inappropriate measurement schema.
- Using a Polarity methodology: Example: Pick one of four statements
that best describes your team... Results are mapped onto an
"Individualism vs. Communitarianism" polarity scale. This process fails to rank order different values,
consequently it is an inappropriate measurement schema.
- Using Forced Choice measurement by presenting four different values
and then selecting only one that is most important fails to create
a rank order relationship between the selected value and other
possible choices.
- The largest area of construct variance happens when people are asked
to manipulate or respond to one set of words that result in a
conclusion that uses a different set of words. For example, presenting
a partial value definition during an individual scoring that
then maps to a different (more worded) value definition when reporting results.
- Static questionnaires have no self-correcting mechanism for testing
the relevancy of the domain of values being used and adjusting the
questions accordingly.
Reliability
If we measure again and again will we get the same results.
One of the challenges of values measurement is to use a ranking schema
to measure a full domain of values in a manner that reduces structural
biasing.
Both Rokeach and Hartman use the ranking process for
values measurement and both use philosophical arguments to define and
classify values in a manner that results in the size of the domain of
measured values being limited to groupings no larger than 18 (Rokeach,1968,1979;
Hartman,1967; Edwards & Davis,1991).
In different research settings Rokeach uses different sets of 18
values (Rokeach,1979).
Hall and Tonna (1978) expanded Rokeach’s domain to
125 values but in turn gave up using the ranking schema. They use a forced
choice methodology. The goal of this
approach, also used by the University Michigan, World
Values Survey, is to make the value set
‘full’ or robust enough to use cross culturally.
Life Journey Map® methodology
resolves this conflict by using interactive technology to randomly present
a large domain of values (130) that are rank ordered in sets of 4. Two
complete ranking rounds (steps) are completed. This methodology generates
both a process measure and a content measure.
Process Measure: 'Mental Model' refers to a analogical (Gentner
& eds.,1983,2001, Johnson-Laird,
1983) manner of decision making. In the ranking process
this results in different patterns of scoring; i.e. ranking one value over
and over again as top priority vs. ranking many different values as
priority. Consequently we can correlate the first step and second step
scores according to descending magnitude of ranking. Mental Model-rank
score correlation = .91. This is a measure of our cognitive
unconscious and demonstrates the reliability of Life Journey Map®
values measurement methodology.
Content Measure: Item measurement refers to specific value
ranking score assigned in each step. The between step Item Correlation
= .45. This is not a particularly high correlation but
understandable given the random presentation of the values. This random
presentation skews the scoring when an important value ends up at the end
of the presentation list disallowing the accumulation of ranking points.
The assumption is that the more ranking steps completed the less the
influence of randomization and the stabilization of the rank order of
specific values.
Preliminary measures indicate that what is reliable over a
period of time is the person's set of Field of Meaning Values.
This is supported by the 71% High Degree of Fit score reported on
top 6 priority values, which are a subset of the individual's Field of
Meaning.
Copyright by Scott Bristol, 2002-2005