You are on page 1of 10

Devasya Jarrett PSYC 6032 Week of January 30

The History of Greek Philosophers- Plato vs Aristotle


ABSTRACT. Philosophy continues to influence numerous disciplines; namely the sciences. At the chief
of these philosophers are Plato and Aristotle; for their opposing views in unraveling the first principle of
rational thought and knowledge. The most accepted view for psychology is Aristotle. Acknowledging the
role of nature via sensual experiences in the creation of knowledge; profoundly explains active reasoning
as the highest form of thinking. Since sexual experiences are observations that act as the backbone for
explaining nature's purpose. Opposing this, Plato noted that sensual experiences are hindrances to
achieving the purest form of reasoning 'the good'. Sensual experiences considered impure and has
corrupted the pure knowledge innately within our souls. Hence, the goal to free the soul through
introspection and suppressing sensual experiences that enable morality has removed empirical
observation. Ultimately, thwarting the development of many disciplines.
Devasya Jarrett - 620076173 PSYC 6032 Week of February 4

Mitchell (1997)- Quantitative Science and the definition of Measurement in Psychology


ABSTRACT. Psychologist for half a century have been troubled with a systemically sustained
methodological thought disorder arising from their inability to cognitively acknowledge and adhere to
methodological facts of quantitative research. This inability is deemed systematic because previous
quantitative psychologist namely Fechner and Steven both blinded but resolute to support their claims
through numerical representation and relativist subjectivism. While other psychologist- Spearman- driven
by practicalism have neglected measurement as one of tenets of scientific investigation. Specifically
ignoring scientific task and apply only instrumental task to suite societies perceived need for measuring
psychological attributes. Thus, psychology has been developing based on a separate definition of
measurement that lacks the category of quantity- Holder (1901)- and its supporting philosophical
knowledge; making it distinct from other disciplines of science.

Borsboom and Mellebergh (2004) -Why Psychometrics is not Pathological


ABSTRACT. This is a critical analysis of Mitchel (2009) work on the pathological state of
psychometrics as a science. The researchers do not agree with this view because Mitchell's rational that
additive conjoint measurement system as an alternative to psychometrics is flawed. The logics is flawed
because Mitchell supports classical test theory but his alternative approach to measurement supports Item
Response Theory which is testable to empirical data. The latter theory is used by Rasch model which is
ultimately the same approach as Mitchell's alternative; however, Mitchel vigorously critiques this model.
I do not really understand this paper.

Schmittmann et al (2013) - Deconstructing the construct


ABSTRACT. In developing a construct for psychology there exist a clear conceptual relationship
between the psychological attribute and the observed variables; expressed in reflective and formative
models. However, the current researchers highlight the problems with these two models and a need for a
refined third model known as networks approach which directly relates to the observable. Some of the
main problems with the previous view on conceptualizing is that the causal relations gained are affected
by time (which precedes the observable or the construct?), poor articulation of relationship between
construct and observables as a process (what made the relationship work?) and the difference in treatment
between observables (valuable causal relations all around). Thus, the network approach proposes that
viewing observables as a part of the construct (a network of variables) that ultimately helps to find
meaningful individual differences with respect to the construct. This helps to eliminate the
aforementioned problems via an understanding that indicators (observables) function autonomously by
connecting causally in the process (autonomous causal entity). This approach is based on the dynamic
systems theory where attractors and state transitions integrate variables to arrive at an equilibrium.
Therefore, the network model presents an adequate approach for conceptualizing psychological constructs
as oppose to the intuition which was once used.
Devasya Jarrett - 620076173 PSYC 6032 Week of February 10

Coombs, et al (1954) - Views on Mathematical Models and Measurement Theory


ABSTRACT. The relationship between mathematical models and scientific measurement was
explained through the two routes used to make draw conclusion s about the real world. The first
route involves a logical argument where a model is mapped into an abstraction, mathematical
system, mathematical argument, theoretical interpretation and a physical conclusion respectively
that is applicable to the real world. This approach only satisfies internal criteria before it
becomes a theory/ prediction about the real world. On the other hand, the second route was
experimentation by means of observation on the real world. It starts with an assertion,
experimental design, experimentation, observation, then statistical interpretation that arrived at a
conclusion about the physical world. The conclusion is considered a form of theory satisfying
internal and external criteria. Furthermore, mathematical models incorporated numerical
operations and scale of measurement set forth by Stevens. An example of the scales relevance to
the two routes of mathematical systems was explained using the partial order vector space,
chains, and other numeric but theoretical terms such as reflexive, transitive.

Stevens (1946)- On the Theory of Scales of Measurement


ABSTRACT. Stevens conceptualized a definition for measurement which is the assignment of
numbers to things to represent facts and conventions about them. He later, proposed the scale of
measurement by which all important information for the quantifying a measure can be gained.
The scales were nominal (unrestricted number assignment), ordinal (order preserving), interval
(quantitative form) and ratio (all measures applicable). His theory was explained in the context
of isomorphism where physical properties in numerical series and empirical operations are
combined to in a model to represent some aspect of an empirical relational systems. Therefore,
this new knowledge of assigning scales to specific properties in order to measure them removed
some of the doubt and discussion surrounding the measurement and quantification of sensation
of loudness. Since the scales combined both the empirical operations and the formal properties of
a physical attribute. This clarity in measurement highlighted the two forms of measurement;
derived (math function of certain magnitude) and fundamental (satisfy criterion of additive).
Hence future questions pertaining to what kind of measurement and the rules that govern each
form of measurement arises.
Fraser (1980)- Measurement in Psychology
ABSTRACT. This paper provides a critical review of the history of measurement that is based in
the philosophy stemming from Plato's ideal view of the abstract purest form existing outside of
nature. To Holder (1901) measurability condition of identifying the order and additive function
of quantity and making a new axiom quantity. In addition, Campbell noted that to produce
quantity as a mathematical equation the properties of magnitude must be carry out the empirical
function of additive. Performing the function is extensive of a fundamental measurement and no
being able is intensive of a derived measurement. The later Ellis (1906) opposed because
magnitudes of a property are distinct hence, they cannot be added to create a new quantity.
Consequently, dominant approaches such as classical measurement based on Stevens definition
and scale of measurement. This posits that isomorphic conditions involving empirical system and
scale value are needed to achieve and proceed with measurement. However, axiomatic approach
of measurement disagreed because theory is the basis upon which measurement is included and
an integral part of empiricism/ experimentation. To axiomatic approach measurement is the
construction of a mathematical model of some property of the world that is assessed via
assignment of number assumptions for measurability. This approach Fraser notes provides
compelling evidence for scale values and variations; however, future research on scale
identification needs to be done.
Devasya Jarrett - 620076173 PSYC 6032 Week of February 19

Lord (1953)- On the Statistical Treatment of Football Numbers


ABSTRACT. A psychometrics professor has taught over the years that ordinal and cardinal
numbers cannot do such; his prime example was test scores. Cardinal numbers are considered
counting numbers and ordinal numbers as positioning numbers. However, the professor loved to
calculate numbers specifically test scores; contradicting his teachings he retired. Given the
opportunity to work directly with numbers (cardinal numbers on a football jersey); the professor
created a machine to assign two-digit numbers on shirts to identify each player on a football
field. Machine tampering and players assigning value to the numbers resulted in a problem he
could not fix. Requesting help a statistician calculated the exact probability of the machine
assigning a high number which caused the problem. He did this by assuming that the football
numbers acted like natural numbers. The professor was dumbfounded by the new revelation that
cardinal numbers in his case 'football numbers' can perform numerical operations such as
multiplication, division, addition and subtraction. Calculating cardinal numbers and test scores
became a task the professor did openly thereafter.

Response to Lord (1954)


ABSTRACT. In support of Lord's football numbers Behan and Behan noted that numbers are
used as names of qualities of objects because they give information about the quality measured.
Therefore, scales of measurement types are determined by the empirical operation and the
appropriate numerical operator chosen then tells the purpose of the number. In addition, Bennett
explained that numbers possessing a cardinal axiom when manipulated by measurement produce
a meaningless test of significance. While Lord himself further explained 'football numbers' using
another example via a game of sample roulette. Here Lord agreed that testing nominal numbers
would lead to an illogical conclusion when the sample is randomly drawn. Hence, care must be
taken when interpreting results of arithmetic operations upon nominal and ordinal numbers.
Sholten and Borsboom (2009)-
ABSTRACT. The theory of admissible statistics by Stevens ensures that conclusions are not
dependent on numerical ale that is arbitrarily chosen to encode data. Hence, statistical test should
result in invariant transformation on all levels of measurement. Lord's 'football numbers' is the
strongest counter argument for admissible theory. When reanalyzed football numbers cannot be
uniquely nominal level only but also interval level to include bias from the machine. Machine
bias was explained by low number football players received. A hypothesis created by the
statistician explained the calculative aspect of football numbers. Thus, making them interval and
observing Steven theory of admissible statistics. This is in accordance with the representational
measurement theory where the level of measurement is determined by the structure of the
property of interest and the relation to numerical assignment bears on the structure. However, the
more important point is that future researchers consider the validity of inferences made leading
to conclusions.

Burke (1953)
ABSTRACT.
Devasya Jarrett - 620076173 PSYC 6032 Week of February 26

Gaito (1980)- Measurement Scales and Statistics


ABSTRACT. A linear flow of the misconception of measurement and statistics was presents for those in
support of Steven scale (nominal and ordinal) leading to non-parametric test and (interval and ratio)
parametric test. Chief among the opposition where those who recognized that scales of measurement
could not be pinned to a statistical test for example t test (Baker et al, 1996) and Gaito (1960) that interval
was not needed for ANOVA test. While Kaiser (1960) and Anderson (1961) noted that scale of
measurement is irrelevant to statistical hypothesis and unimportant in selecting a parametric and non-
parametric test. Therefore, measurement scales being requirements for statistical test is incorrect because
assumptions to ensure measurement concerns for proper statistical conclusions are the real requirement.
Hence, math test does not use scale properties as requirements for statistical test; it is a flawed belief of
statistical theory and measurement. Stevens only value was developing scales of measurement to help the
measurement theory not the statistical theory.

Gaito (1986)- Some Issues in Measurement- Statistics Controversy


ABSTRACT. A distinction between the views of anti-measurement (AM) group and the pro –
measurement (PM) group based on Stevens fusion of measurement and statistical testing is still an area of
controversy. The AM group note that anything goes with measurement and stats because there is not
enough evidence to show that scales of measurement are applicable to all statistical test or even relevant
in experimental design. Instead, they believe that statistical analysis does not equal to experimental design
because statistical analysis is just a stage in the experimentation process of conducting the experiment.
This stage involves testing of hypothesis and rejecting or failing to reject a null hypothesis. These
conclusions are statistical in nature; thus, lacks reference to measurement properties or considerations
which takes away from the associated meaning in interpretation stage. On the other hand, the PM group
noted that the overall experimental design stages are important when measurement properties and
statistical analysis are included together. That is measurement scales properties are determined by specific
statistical test with the notion of one scale or level of analysis related to one test. However, more than one
levels of scale measurement can be assigned to a statistical test and be tested hypothetically as such, or
two or more procedures have different scales but same result, scales that are adjacent scale properties can
predict the same results when tested and scales that are far apart (nominal and interval) can produce same
results (normal approximation and binomial distribution). Another flaw of the PM group is the incorrect
understanding of statistical test and what they require to produce the end result; for example, Pearson's r
is proposed to be interval ratio but in the definition of Pearson's it is actually a ranking of the values from
an interval results. Also, the PM group does not recognize that permissible transformations can happen to
any scale and mathematical operations can be performed by all scales. Consequently, not understanding
the statistical and empirical conclusion validity gained when focusing on the overall research effort
hinders scientific nature of research. Hence, AM group supports that measurement and meaning can enter
into same stages of experimental design but not into statistical analysis.
Townsend and Abby (1984) Measurement Scales and Statistics: The Misconception Misconceived
ABSTRACT.
Devasya Jarrett - 620076173 PSYC 6032 Week of March 3

Stine (1989) Meaningful Inference


Stevens core assumptions is that measurement structures are categories or classified; they are
invariant in numerical structures and are useful for science because of their meaningful
interpretation. Numerous critics of statistic and measurement relation identify the diff in theory
concerning the debate (Michel, 1986) - representational, operation, classical. But meaningful
operations cannot be obtained in a laboratory and measurement and scientific theory not status,
operational and classical theory respectively. Similarly, invariance cannot be applied to ordinal
scale or stats (Anderson). Meanwhile the assumption that all scales fit normal distribution and
are interval scale is not so because ANOVA (Gaito 1959) and Jenson (1980) does not fit those
assumptions. However, representational theory linked with Campbells axiomatic approach
explains his assumptions into a logical argument. That is statistical analysis are unchangeable
under scales of valid transformation, so that they can be interpreted empirically.

Michell (1986)- Measurement and Statistics: Clash Paradigms


A comparison between the three dominant theories or perspectives on measurement
(representational, operational and classical theory) are assessed based on their corresponding
implication on statistics test. The representational theory through Townsend and Ashby noted the
role numbers play in explaining qualitative data as empirical relations that can be equivalent,
ordered, applied to different admissible scale transformations and be concatenated based on
similar transformations. These transformations are permissible to scale types and are meaningful
or meaningless statements. However, scale free statements are preferred because it is
scientifically useful since all statements implies truth about a sample (meaningless) and abut an
artefact (meaningful). Similarly, a shift from scale specific measurement statements to scale free
statements using meaningfulness criterion improves the ability of empirical
structures/observations to form conclusion that are valid scientific analysis (Adam et al 1965).
And doing so by using mathematical logic arguments to draw empirical arguments (Campbell
and Russel 1920). While Operational theory states that measurement process is simply an
operation that produces number or based on some kind of an object to be measured. That is the
number assigned is the outcome of the operation. Operationalist believes that the data on which
measurement is based is inherently (by nature) numerical; because numbers are the end product
produced NOT acting as convenience as representationalist say. Thus, operationalist seek to
discover new quantitative relations among numbers assigned and uses the scale types to imply
information as needed in measurement. Also, there is no restriction on scale types and statistical
test because the result from test are the end and not a stage in the experimental process towards
scale free conclusions. These principles in operationalist are expressed in test scores and rating
scales. However, some psychologist does not agree because rating scale reflect an underlying
psychological variable that needs meaning applied to it from the calculations. Therefore, the
classical theorist noted that measurement is an assessment of quantity and how much of a given
attribute some object possesses. The structure of the attribute helps to determine if

You might also like