14 Quantitative Methods - Necessary Condition Analysis Flashcards
What is the Necessary Condition Analysis
The distinction between necessary and sufficient conditions is an old one, but it has been methodologically somewhat challenging to identify necessary conditions. A novel technique, known as Necessary Condition Analysis (NCA) allows us to study if a condition is necessary for an outcome to occur. The method has not yet been applied in the field of political science but has been employed in several other research fields.
What are the advantages of the NCA?
(1) It can easily be employed as a complementary method to, for example, regressions in quantitative papers and can lead to potentially fruitful insights.
(2) It requires us to focus more on the visual inspection of the data – something that is currently undervalued in research practice.
(3) It encourages us to focus on the no-data zones, and challenges us to think differently about our data.
Questions on Tynan, Michael, Marcus Credé and Peter D. Harms (2020): “Are individual characteristics and behaviors necessary-but-not-sufficient conditions for academic success?
(1) What is the research question?
(2) What are the main results of the study?
(1) Are individual characteristics and behaviours necessary-but-not-sufficient conditions for academic success?
(2) Results of the NCA indicate that the observed data are consistent with modest necessary-but-notsufficient relations between grades and some of the examined noncognitive abilities, traits, and academic skills. Significance tests of the necessary condition analysis reveal that perseverance, conscientiousness, growth mindset, impulse control, self-esteem, class
attendance, cognitive ability, and high school GPA are possible necessary conditions for academic success in college.
Questions on Dul, Jan (2016): “Necessary Condition Analysis (NCA): Logic and Methodology of ‘Necessary but not Sufficient’ Causality”.
(1) Read the introduction: What are necessary conditions, and how is the study of necessary conditions different in comparison to additive causality models?
This notion of causality can be interpreted as having two distinct logical parts. The first quoted sentence reflects the common understanding of causality, namely, that X is sufficient for Y: ‘‘If X, then Y.’’ Most people tend to understand ‘‘X causes Y’’ in terms of sufficiency. This may be the reason that hypotheses in organizational sciences frequently imply sufficiency. The second quoted sentence reflects that X is necessary for Y: ‘‘No Y without X.’’ Such hypotheses are rarely formulated and tested in organizational sciences.
Although scholars often confuse necessity and sufficiency (Chung, 1969; Goertz & Starr, 2003), the two are totally different.1 A necessary cause allows an outcome to exist; without the necessary cause, the outcome will not exist. A sufficient cause ensures that the outcome exists; it produces the outcome. A student who wants to be admitted to a U.S. graduate school (the outcome) needs to have a high score on the Graduate Record Examinations (GRE) test: An adequate GRE score is necessary for the outcome. Necessary causes are not automatically sufficient. An adequate GRE score is not sufficient for admission because also other admission requirements play a role (e.g., the student’s motivation letter). However, if the student’s GRE score is too low, there is guaranteed failure, independently of the student’s performance on the other requirements. Therefore, a necessary cause is a constraint, a barrier, an obstacle, a bottleneck that must be managed to allow a desired outcome to exist. Prevention of guaranteed failure and increased probability of success are core constituents of the ‘‘necessary but not sufficient’’ logic of ‘‘X causes Y.’’
What are the dichotomous, discrete and continous necessary conditions?
There are three types of necessary conditions, dichotomous (ikiye bölünen), discrete, and continuous. In the aforementioned examples, both the condition (GRE score, fuel, trust) and the outcome (admission, moving car, solid financial system) can have only two values: absent or present. This is the dichotomous logic of necessary conditions, which is part of classic binary (or in mathematics, Boolean) logic. However, many real-life situations are not inherently dichotomous (although they can be dichotomized). In the discrete situation, the necessary condition and the outcome can have more than two values (e.g., low, medium, high), and in the continuous situation, any value is possible. The contingency matrix is a common way to present necessary conditions.
How can “Necessary Condition Analysis” be methodologised?
the general methodology necessary condition analysis is developed for determining necessary (but not sufficient) conditions. This methodology consists of two main parts: (1) determining ceiling lines and the corresponding bottleneck tables and (2) calculating several parameters such as accuracy of the ceiling line, effect size of the necessary condition, and necessity inefficiency.
What is the “ceiling line” and what is the “ceiling zone”?
The starting point for the necessary condition analysis is a scatter plot of data using a Cartesian coordinate system, which plots X (the determinant and potential necessary condition) against Y (the outcome) for each case.
To summarize, the ceiling line represents the minimum level of a necessary condition required for the outcome to be possible, while the ceiling zone represents the optimal range of values above the ceiling line where the outcome is most likely to occur.
There are two techniques to draw the ceiling line, namely “ceiling envelopment” and “ceiling regression”. Explain the differences between these methods.
The ceiling line represents the minimum level of a necessary condition required for the outcome to be possible, while the ceiling zone represents the optimal range of values above the ceiling line where the outcome is most likely to occur.
the ceiling line represents the minimum level of a necessary condition required for the outcome to be possible, while the ceiling zone represents the optimal range of values above the ceiling line where the outcome is most likely to occur.
The accuracy of a ceiling line is defined as the number of observations that are on or below the ceiling line divided by the total number of observations, multiplied by 100%. Then, by
definition, the accuracy for CE-VRS, CE-FDH, and COLS is 100%, and for the other techniques, the accuracy can be below 100%.
What is “effect size” and how is it calculated?
Perhaps in nearly all scatterplots, an empty space can be found in the upper left corner, which may indicate the existence of a necessary condition if this makes sense theoretically. The question then is: Is the necessary condition large enough to be taken seriously? Therefore, there is a need to formulate an effect size measure. An effect size is a ‘‘quantitative reflection of the magnitude of some phenomenon that is used for the purpose of addressing a question of interest’’ (Kelley & Preacher, 2012, p. 140). Applied to necessary conditions, the effect size should represent how much a given value of the necessary condition Xc constrains Y.
The effect size can be expressed as follows: d = C/S, where d is the effect size, C is the size of the ceiling zone, and S is the scope. Hence, d is the proportion of the scope above the ceiling. The scope can be determined in two different ways: theoretically, based on theoretically expected minimum and maximum values of X and Y, and empirically, based on observed minimum and maximum values of X and Y. Thus: S = (Xmax – Xmin)*(Ymax – Ymin).
What are the limitations of NCA?
- NCA cannot solve the problem of ‘‘observational data cannot guarantee causality.’’ Observing a data pattern that is consistent with the causal hypothesis is not evidence of a causal connection
In case you are interested in the method, you should read the section “recommendations for applying NCA”.
NCA can be applied in any branch of organizational research (and elsewhere) in which theoretical necessary (but not sufficient) statements are (or can be) formulated.
What are the differences between NCA and QCA?
NCA sounds
a bit like QCA, but is really not. For example:
(1) In QCA, we study several conditions and their configurations. In NCA, the focus
is on one condition at the same.
(2) While QCA allows the testing of necessity relationships, the focus is rather on
sufficiency.
(3) QCA identifies necessary conditions in kind (existence of X is necessary for Y).
NCA allows us to identify also necessary conditions in degree (a certain degree of
X is necessary for certain degree of Y).
(4) QCA requires us to calibrate the variables (always a subjective decision), NCA
uses data as it is.
