Research Governance in Public University: Proposal of a Composite Index

Abstract

This work aims to elaborate a governance composite index for research laboratories in public university. This index is composed by an indicator of responsible liberalism associating effectiveness and ethics, to which are added an organizational management indicator and a strategic management indicator. To achieve the above aim, several methods are used, such as adjusted data envelopment analysis and geometric mean to aggregate indicators to calculate the composite index, Vigier index to compute responsible liberalism indicator, and tools to measure the validity and the reliability of indicators. The findings show that the developed index can be applied in any context.

Keywords

research governance composite index responsible liberalism indicator adjusted data envelopment analysis Vigier index validity and reliability of indicators

Introduction

Governance has become essential for the development of organizations because it ameliorates their relations with their various stakeholders, including their public and private governments, and therefore improves their performance (Ysa et al., 2014). The university’s research laboratory does not escape this rule. Since governance is multidimensional in nature, the indicator of its measurement must be composite.

For the development of the composite index in organizations, including research laboratories in the university, several methods can be used, such as institutional methods (Kaufmann et al., 2007, 2010), parametric methods, such as stochastic frontier analysis (Khiari et al., 2007) and discriminant analysis (El Kadiri Boutchich, 2020a), and index methods (Coccia, 2004).

However, these methods have some drawbacks. First, the institutional methods involve certain biases. Second, parametric methods require several tests and are difficult to specify, manipulate, and interpret when they take a nonlinear form (El Kadiri Boutchich, 2020b). Finally, the index methods are approximate and do not make it possible to identify the main components of the composite index.

The method used in this work enables to face the drawbacks of the aforementioned methods. In addition, it adopts a normative approach that has become scarce in the social sciences (Ribes & Baker, 2007). Furthermore, it combines qualitative and quantitative indicators for more relevance of a composite indicator (El Kadiri Boutchich, 2019).

In addition, this work is original because it uses a method called “Adjusted Data Envelopment Analysis (Adjusted DEA),” which allows processing only the outputs. Moreover, it uses exploratory factor analysis instead of confirmatory one to check the validity and reliability of the indicators because it is compatible with the nature of this research, which is exploratory. Finally, this research is equally original, insofar as it adequately synthesized two qualitative variables to calculate the responsible liberalism indicator.

Likewise, currently there is no reference framework for the development of governance indicators within Moroccan universities, just as at present, there is no catalog of current governance practices even within European universities. In the same way, the adjusted composite index of research excellence proposed by the European Commission does not include any governance indicator within it (Arregui-Pabollet et al., 2018). Thus, this work enables increasing the debate about implantation of governance models within universities in general and in research laboratories in particular.

Here, it is necessary to indicate that scientific research constitutes the real human capital of the university, and therefore of the region and country where the university is located (El Kadiri Boutchich, 2020d). Likewise, governance is essential to improve human capital and therefore efficiency in university including its research laboratories (Bayyurta & Yilmaz, 2012).

Finally, the public university was chosen for this research because governance problems arise more within it than the private university (Ddungu & Edopu, 2016).

In terms of its presentation, this manuscript is articulated, in addition to the Introduction, around five parts: Conceptual Review, Method, Results, Discussion, and Conclusion.

Conceptual Review Around Composite Index and Academic Research

Composite indexes are a special type of measure designed to quantify phenomena that are multidimensional, such as governance. The development of a composite index passes several stages. First, it is necessary to select the relevant elementary indicators. Then, if the number of these indicators is very high, it is advisable to reduce them through factor analysis (Babcicky, 2013). After, it is suitable to weigh them and to aggregate them. Weighing techniques can be normative based on the experience of the designer, numerical, or statistical. As for aggregation techniques, they can be additive or multiplicative (Muriithi et al., 2015). Finally, standardization of composite indexes is essential to allow adequate comparison between the entities concerned.

Governance, including research, could be measured according to two types of methods: simple and less simple. Simple methods comprise mean of weighted or non-weighted indicators, rating methods, and indexing ones. The less simple methods include unobserved components methodology (UCM) elaborated by the World Bank (Kaufmann et al., 2007, 2010), which is effective when the information sources are various, the benefit-of-doubt model, which allows weighting elementary indicators and standardizing the synthetic indicator (Streimikiene et al., 2016).

The less simple methods also use multi-criteria approaches, such as value, utility, distance functions (El Gibari et al., 2019) and best worst method (Salimi & Rezaei, 2016), and the synthetic indicator proposed by Melyn and Moesen (1991), based on linear programming with cumulative constraints. Furthermore, they employ discriminant analysis (El Kadiri Boutchich, 2020a) and stochastic frontier analysis (Khiari et al., 2007).

Other methods could be applied in composite index area, notably for indicator weightings, such as budget allocation process, analytic hierarchy process, and conjoint analysis (OECD, 2008).

Specifically, concerning the governance of research in university, it incorporates in particular accountability, ethics, and efficiency (Shaw et al., 2005). The research governance also includes responsibility, innovation, and integrity (Arregui-Pabollet et al., 2018). In addition, it contains transparency and participation (Abdeldayem & Aldulaimi, 2018). Equally, it integrates cultural diversity, equity, and research internationalization (Welch, 2020). Moreover, laboratory governance comprises their organizational, strategic, ethical, and financial aspects (High Council for Evaluation of Research and Higher Education, 2015, 2020).

Furthermore, as part of a submission to Research Excellence Framework by the Edinburgh University to assess governance within it, in 2021, governance will include in particular strategy and commercialization of research, and innovation (University of Edinburgh, 2020). Finally, from data of Research Assessment Exercise 2001, 2008, and Research Excellence Framework 2014, research governance encompasses accountability, research funding, and laboratory outcomes (Oancea, 2019). Finally, from Star Metrics’ perspective, governance refers to the return on investment in research (Adam et al., 2018).

It should be indicated that governance of research in university, according to the above works, uses several methods, such as rating methods, economic analysis, including the cost-benefit one, peer review, and case study.

Regarding now the governance composite index, it seems that it is not practiced in the laboratories of the university. That is, why research governance is often integrated into an indicator of global governance of university (Arregui-Pabollet et al., 2018), or even into an index of overall performance of the university (Kwiek, 2015) or into an index of overall research performance within university (Ferretti et al., 2018).

These indexes, which are composite, incorporating an indicator of governance, used notably rating technique, unweighted arithmetic mean, and unweighted geometric mean.

Thus, in the absence of a composite index within university’ laboratories, it is legitimate that they adopt the methods aforementioned notably these of managerial nature. This assertion is based on the fact that research laboratories today are considered to be enterprises or industrial organizations (El Kadiri Boutchich, 2020b). This assertion is equally legitimized because it is the governance entrepreneurial approach that is most coveted by research laboratories, compared to the state-centered approach and autonomy approach (Abdeldayem & Aldulaimi, 2018).

Methodology

It includes problematic, epistemological stance approach, methods used, and implementation related to the composite index developed in this work.

Problematic

Nowadays, the works of design and development of measuring instruments in economics and in management are relatively weak with regard to the related conceptual production. Moreover, there is a gap between the instrument and the concept it conveys, especially if it is based on an open artifact (Martineau, 2017). Likewise, the public sector instruments, the subject of this study, do not measure the completeness of performance (Savall & Zardet, 2011).

Finally, there is no governance composite index in laboratories of public university. This work seeks thus to fill these gaps.

In view of the above, the question posed by this research is the following: How to adequately determine the composite index of governance within the research laboratories of the public university while ensuring the appropriateness of weightings, aggregation, and validity and reliability of indicators?

Epistemological Stance and Approach

This work adopts a positivist reasoning of an exploratory nature, which avoids the formulation of hypotheses and the need to confirm or invalidate them (El Kadiri Boutchich, 2020c). To ensure the objectivity of the research and not to influence the respondents, a questionnaire is administered to a sample of 21 research laboratories in different disciplines, belonging to Mohammed First University of Oujda-Morocco during 2019 to provide data of 2018. These laboratories are the most structured and therefore can complete the questionnaire without problems and without any bias. In addition, the questionnaires were completed by laboratory directors.

For governance measurement, partnership, cognitive, and ethical approaches are combined, as long as the research laboratories must satisfy ethically their stakeholders and adopt innovation and creativity to improve their outputs.

Method

Four methods are used in this work: Adjusted DEA, Vigier index, tools to measure the validity of the indicators, and tools to highlight their reliability.

Adjusted data envelopment analysis

It is a transformed form of DEA Standard where only the outputs are taken into account to give them an optimal weight and to calculate the composite index because a composite index must contain only outputs. At this level, it should be remembered that the adjusted composite index of research excellence proposed by the European Commission has been revised to eliminate the inputs and keep only the outputs in this index (Vertesy, 2018). The transformed program is as follows (El Kadiri Boutchich, 2020b):

\begin{array}{l} {Min}_{θ, λ} θ subject to : \\ \begin{array}{l} - y_{r k} + \sum_{j = 1}^{n} λ_{j} y_{rj} \geq 0 : I n e q u a l i t y (1) \\ θ - \sum_{j = 1}^{n} λ_{j} \geq 0 : I n e q u a l i t y (2) \\ λ_{j} \geq 0 \end{array}} P r o g r a m 1 \end{array}

where is the quantity of the output r relative to the entity k, n is the number of entities processed in the program, $λ$ is a vector of constants, and $θ$ is a scalar between 0 and 1, representing the composite indicator.

For the users who do not manage to pose the linear program above, it is possible to multiply the inequality (2) of the program 1 by a constant A; this gives rise to the program 2 below:

\begin{array}{l} {Min}_{θ, λ} θ subject to : \\ - y_{r k} + \sum_{j = 1}^{n} λ_{j} y_{rj} \geq 0 : I n e q u a l i t y (1) \\ θ A - \sum_{j = 1}^{n} λ_{j} A \geq 0 : I n e q u a l i t y (2) \\ λ_{j} \geq 0 \end{array}} P r o g r a m 2

Program 2 looks like the standard DEA program. To solve it, it suffices to retain as input a constant value equal to A and to use standard DEA software (some DEA software are mentioned in conclusion).

Adjusted DEA is preferred to a geometric mean as it allows having optimal indicator weights and a non-zero aggregated index even with null some data.

Vigier index

The Vigier index is used to determine an index according to a global vision such as that of total quality. It is calculated according to the following steps (El Kadiri Boutchich, 2019): $S_{Y} = \exp (- {| U |}^{n}), with U = \frac{2 y - (y \max + y \min)}{(y \max - y \min)};$ where y is the score attributed by a given laboratory to effectiveness-ethics system, y max and y min are the lower and upper limits, and n is the degree of sensitivity of the Vigier index. So, $n = \frac{{\ln (\ln (1 / S}_{y}))}{\ln (| U |)};$ where ln is Napierian logarithm.

This index is used to calculate the responsible liberalism index, which combines ethics and effectiveness. A system comprising these two dimensions is chosen as they fit well with approaches employed in this work. Ethics ensures distributive justice between laboratories and their stakeholders, while effectiveness ensures their perennity that allows innovation to take place. Thus, each laboratory must assign a score, qualitatively to this system according to the following scale.

It is assumed that the optimal solution belongs to the class between 40 and 60, where there is a perfect balance between ethics and effectiveness. Then, it is necessary to define lower and upper limits. These limits depend on the rigor with which the laboratory intends to control its system of effectiveness-ethics. This control will be more rigorous as these limits will be close to the class comprising the optimal solution.

In fact, the lower limit is set at 20 (below this, the control system is ineffective) and the upper limit is set at 80 (beyond this, the control system is unethical). Likewise, it is accepted that when the score is equal to 60, the quality index is equal to 0.9. In fact, statistically, 0.9 is a quite significant threshold. It is also possible to associate 0.9 with 40 because the quality index is a symmetrical function. From these data, it is possible to determine the Vigier index formula by determining the values of U, n, and Sy. $U = \frac{2 y - (y \max + y \min)}{(y \max - y \min)}$ $= \frac{2 \times 60 - (80 + 20)}{(80 - 20)} = \frac{1}{3}$ ; $n = \frac{\ln (\ln (1 / 0, 9))}{\ln (| 1 / 3 |)}$ = 2,048. Then, $S_{Y} = \exp (- {| U |}^{2, 048}) .$ when laboratories assign scores to the ethics-effectiveness system, y is the score given by each laboratory to this system.

Tools to measure the validity and reliability of the composite index

The validity is divided into two typologies, internal/external and convergent/discriminant. Internal validity refers to the fact that the instrument used actually measures what one wants to measure. Palpably, to ensure internal validity, one must use at least two different instruments that lead to the same result. As for external validity, it requires that the data collected from the respondents be accurate and that the sampling does not introduce any bias (Muriithi et al., 2015). Regarding the second typology, convergent validity requires items that measure a common underlying construct, while discriminant validity assumes an indicator should not be influenced by underlying constructs than the one it is supposed to measure.

For the first typology, the correlation coefficients and sampling relevance indices (Kaiser-Meyer-Olkin [KMO] index and Bartlett’s sphericity test) are used in particular. For the second typology of validity, the most used methods are exploratory factor analysis and confirmatory one. Concerning convergent validity, factors loadings, average variance extracted (AVE) and construct reliability (CR) are retained in particular. As for the discriminant validity, it employs several indicators, such as the following (Cai et al., 2020): chi-square, root mean square error of approximation (RMSEA), root mean square of the residuals (RMSR), comparative fit index (CFI), normed fit index (NFI), and goodness of fit index (GFI).

Some formulas of indicators employed in this analysis are given. These indicators will notably fit with an exploratory factor analysis with maximum likelihood as an extraction method and varimax as the method of rotation. Maximum likelihood is used, as it allows calculating a more validity and reliability indicators than the principal axis factoring method which can also be utilized for exploratory factor analysis. Varimax rotation is retained because it is orthogonal and makes the interpretation of the factors more simple and reliable. These indicators are as follows:

KMO value: It is provided by SPSS (software);

Sphericity test: It is processed by SPSS;

$A V E = \frac{Σ_{1}^{h} λ^{2}}{h}$ ; where λ is the correlation coefficients of the variables with the component concerned; h is the number of variables correlated with the component concerned;

$C R = \frac{{(Σ_{1}^{h} λ)}^{2}}{{(Σ_{1}^{h} λ)}^{2} + Σ_{1}^{h} {(1 - λ)}^{2}}$

χ² (according to the maximum likelihood extraction): It is computed through SPSS.

$R M S E A = \sqrt{\max (\frac{χ^{2} - df}{df (n - 1)}; 0)}$ ; where df is the degrees of freedom of the model and n is the sample size.

Concerning reliability, it allows appreciating coherence and homogeneity of items within each component. To measure it for the composite index, Cronbach’s alpha is used. Cronbach’s alpha = $\frac{p}{p - 1} (1 - \frac{Σ_{1}^{p} {V (Y}_{i})}{V (X)})$ ; where V(Y_i) is the variance of the item Y_i while V(X) is the variance of total scores of all items.

Implementation

For elaborating the standardized composite index of governance, seven items borrowed from conceptual review are retained: participation (participation rate in general assemblies), consultation concretized by the number of general assemblies, and transparency reflected through the presentation or not of the activity and financial report (AFR) of the laboratory. Number of laboratory council meetings, definition of the laboratory strategy, the rate of its implementation, and ethics-effectiveness system are also part of the seven items. About the last item, research ethics comprise notably autonomy, beneficence and non-maleficence, and justice (Council of Europe, 2012). Research ethics also includes social norms and values, compliance with regulations, statutes and institutional policies, rigor and reproducibility of science, and finally, workplace relationships (DuBois & Antes, 2018).

As for effectiveness, it is the realization degree of laboratory scientific and economic outcomes. The aggregation of the seven items into some indicators is carried out through Adjusted DEA, while aggregation of these indicators into a composite index is done through a simple geometric mean to avoid overestimation of the scores. For transparency and strategy definition, the presence of AFR or strategy is counted as 1, while their absence is counted as 0.

Results

The results comprise elementary indicators, calculation of the validity and reliability of these indicators, and the computation of the governance standardized composite index. The elementary indicators are presented in Table 1.

Table 1.

Elementary Indicators.

Laboratory	Participation	Consultation	Transparency	Laboratory council meetings number	Strategy definition	Strategy implementation rate	Ethics-effectiveness system scores
1	50	1	0	2	0	0	75
2	55	2	1	1	0	0	45
3	45	1	0	2	0	0	30
4	50	1	0	4	1	70	65
5	50	2	1	2	0	0	80
6	40	1	0	2	0	0	75
7	50	1	0	4	1	40	75
8	75	2	1	6	1	50	60
9	50	1	0	4	1	0	75
10	50	1	1	4	1	0	60
11	60	2	1	2	0	0	30
12	25	1	0	2	0	0	30
13	75	2	1	6	1	60	60
14	50	1	0	4	1	0	70
15	50	1	0	4	1	0	65
16	75	2	1	6	1	65	75
17	60	2	1	4	1	80	45
18	25	1	0	2	0	0	40
19	75	2	1	6	1	50	75
20	50	1	0	4	1	0	75
21	50	1	1	4	1	80	30
Mean	52.857	1.381	0.476	3.571	0.619	23.571

Source. Own elaboration.

Participation, transparency, and strategic implementation rate are weak. About effectiveness-ethics system, laboratories, which reconcile ethical and effectiveness, well (scores between 40 and 60) represent only 28.57%. Laboratories, which favor ethics (scores below 40), represent only 19.05%, while the percentage of those who aim for effectiveness is 52.38 (scores above 60). Ethics-effectiveness system scores mean is not calculated because it is not significant due to the double dichotomous dimension of the rating scale of this system.

Internal/External Validity

The goal of internal validity is to show that items require grouping through factor analysis. It is checked by two different instruments to highlight it. These instruments are KMO and Bartlett’s test of sphericity in Table 2. Here, it should be indicated, to appreciate KMO, some researchers retain the seven levels oscillating between unacceptable and marvelous, proposed by Kaiser in 1974 in his article “An index of factorial simplicity.” However, KMO is calculated differently from the factorial simplicity index which uses the quartimax transformation criteria of Carroll, Wrigley, Neuhaus, and Saunders. In fact, most data analysis software retain a value between 0.5 and 0.6 as minimum level of data relevance for factor analysis.

Table 2.

Kaiser-Meyer-Olkin (KMO) and Bartlett’s Test.

KMO measure of sampling adequacy	.718
Bartlett’s test of sphericity	102.482
	21
	.000

Source. SPSS.

KMO being greater than 0.6 and the chi-square relative to the Bartlett’s test being very significant (p = 0), it is possible to affirm that the items are factorizable. Their three-factor classification according to the maximum likelihood method with varimax rotation is consigned in Table 3.

Table 3.

Rotated Factor Matrix.

	Factor
Items	1	2	3
Participation	.522	.720	.286
Consultation	.030	.985	.030
Transparency	.216	.833	−.183
Laboratory council meetings number	.883	.251	.290
Strategy definition	.916	−.024	.234
Strategy implementation	.693	.352	−.153
Ethics-effectiveness system	.167	−.035	.850

The values in bold are the items that are most correlated with one of the three factors.Source. SPSS.

According to Table 3, laboratory council meetings number, strategy definition, and strategy implementation are part of Factor 1, which can be labeled “Strategic Management.” Participation, consultation, and transparency belong to Factor 2, which can be called as “Organizational Management,” while ethics-effectiveness system is linked to Factor 3, which is termed as “Ethics-Effectiveness Management.” This structure is highlighted by Figure 1 below.

Figure 1.

Indicator structure.

Concerning the external validity, the data are presented in the form of numbers whose probability of non-accuracy tends toward 0. In addition, the selection of the sample of 21 does not involve any bias, insofar as only these 21 laboratories are sufficiently structured and could give information required by this research.

Convergent/Discriminant Validity and Reliability

About convergent validity, it is processed with reliability measured through Cronbach’s alpha, as convergence validity and reliability are close together. The indicators of the two are presented in Table 4.

Table 4.

Indicators of Convergent Validity and Reliability.

Indicators (latent variables) with their items	Factor loadings	Average variance extracted (AVE)	Construct reliability (CR)	Cronbach’s alpha
Strategic management		.70	.87	.897
Laboratory council meetings	.883
Number	.916
Strategy definition	.693
Strategy implementation		.73	.89	.849
Organizational management	.720
Participation	.985
Consultation	.833
Transparency		.72	.72	—
Ethics-efficiency management
Ethics-efficiency system	.850

Source. Own calculation from SPSS.

Factor loadings are good because their values exceed .6. Also, the values of AVE are adequate insofar as they exceed the value of .5. The same assertion can be made for CRs, which are greater than .7. Thus, the convergent validity of the indicators is satisfactory.

For reliability, Cronbach’s alpha is employed. Before calculating it for each factor, it is necessary to transform the data related to the indicators to scales. To do this, the data are processed through hierarchical classification analysis, using within groups linkage as cluster method and square of Euclidian distance as interval.

Since Cronbach’s alpha is much higher than .6, reliability of the model is good. Indeed, its acceptance threshold is > .6 for exploratory analysis and > .8 for a confirmatory one (Ghewy, 2010). For the third factor, Cronbach’s alpha could not be calculated because it contains only one item.

About the discriminant validity, the indicators mentioned previously are computed in Table 5. Only one factor is not retained because it does not make it possible to operate the adopted rotation (Varimax), whereas beyond three factors, the number of degrees of freedom (df) becomes negative. This is why only two factors or three factors can be processed.

Table 5.

Indicators of Discriminant Validity.

Factors	Chi-square			RMSEA
Factors	Value (1)	df (2)	(1)/(2)	RMSEA
2	7.73	8	0.97	0
3	1.098	3	0.37	0

Source. Own calculation from SPSS.

Note. RMSEA = root mean square error of approximation.

N.B.: df = ((p − m)² − p − m)/2; p is the number of items (in this case p = 7) and m is the number of factors retained. To examine these assertions above, the following calculations are performed: m = 2, df = ((7 − 2)² − 7 − 2)/2 = 8; m = 3, df = ((7 − 3)² − 7 − 3)/2 = 3; m = 4, df = ((7 − 4)² − 7 − 4)/2 = −1. In this case, model of four factors cannot be treated. In the same vein, it is possible to verify that beyond three factors, df is negative and therefore the model cannot be processed.

As stated by Table 5, the retention of three factors gives better indicators compared with two factors, with a smaller chi-square/df ratio (0.37 versus 0.97) and equal RMSEA of 0. Thus, the discriminant validity level is great with three factors, which give good indicators’ values.

Computation of the Standardized Composite Index of Governance

It goes through two stages. At the first stage, items belonging to each factor are aggregated through Adjusted DEA. At the second stage, the outputs of the first stage cannot yet be considered as outputs to avoid overestimating the scores. Indeed, these outputs have already been optimally weighted and standardized. A second same operation will overestimate their scores. Therefore, the approach of Kao and Liu (2011) is adopted using a simple geometric mean at the second stage of DEA. Thus, the same principle of Kao and Liu is applied in the second stage for indicators’ aggregation. Results are presented in Table 6.

Table 6.

Governance Standardized Composite Index.

Laboratory	Strategic management	Organizational management	Ethics-effectiveness management	Governance standardized composite index
1	0.333	0.667	0.50	0.481
2	0.167	1	0.97	0.545
3	0.333	0.6	0.65	0.506
4	1	0.667	0.79	0.808
5	0.333	1	0.37	0.498
6	0.333	0.533	0.50	0.446
7	1	0.667	0.50	0.693
8	1	1	0.90	0.965
9	1	0.667	0.50	0.693
10	1	1	0.90	0.965
11	0.333	1	0.65	0.600
12	0.333	0.5	0.65	0.477
13	1	1	0.90	0.965
14	1	0.667	0.65	0.757
15	1	0.667	0.79	0.808
16	1	1	0.50	0.794
17	1	1	0.97	0.990
18	0.333	0.5	0.90	0.531
19	1	1	0.50	0.794
20	1	0.667	0.50	0.693
21	1	1	0.65	0.866
M	0.738	0.8	0.68	0.708

Source. Own calculation.

The ethics-effectiveness system score, which is assessed qualitatively, is transformed into a Vigier index that is purely quantitative and standardized, situated between 0 and 1, through U and S_Y formulas previously presented. The mean Vigier index, which combines effectiveness and ethics, is acceptable (0.68). Likewise, the aggregation of the three governance indicators gives a mean of composite index, which is good enough (0.708). Moreover, according to the relative evaluation, Laboratory 17 ranks first, while Laboratory 6 ranks last in terms of governance practice.

Among the three factors, organizational management is the most effective, while ethics-effectiveness management is the less effective. However, these results should be interpreted with prudence, insofar as the scores of 1 are attributed to the benchmarks, while the scores of the other laboratories are calculated on the basis of benchmark scores. These are therefore relative scores and not absolute ones.

Discussion

Regarding Table 1, the element that deserves to be clarified is the weakness of reconciliation between ethics and efficiency, with adequate reconciliation rate of 28.57% only. This weakness does not seem to be a problem related to research in Morocco. Indeed, this problem was mentioned on theoretical and practical levels. At the theoretical level, it is manifested through normative economic theory, which superficially integrates the ethical variable in the economic effectiveness while affecting the research related to economics of education (Wight, 2017). On the practical level, effectiveness in applied research encounters an array of ethics challenges because of ethics and regulatory issues that have emerged currently (Sugarman, 2016).

To surmount this problem in academic research, several solutions are proposed, such as combining normative approach and empirical one (Alzola, 2011) or applying participatory research, whose aim is the reconciliation of community needs and research effectiveness. However, these goals are difficult to achieve because of divergent research practices (McCracken, 2020), thus requiring innovative research designs (Allen et al., 2017). At this level, the raw indicator of responsible liberalism proposed by this work, in terms of an appropriate reconciliation rate of effectiveness and ethics, could be an adequate basis for research decisions and designs.

However, the discussion of the results of Table 6 will not be of great interest, as they are relative and result from a method, which is not applied elsewhere in the field of the governance. In addition, there is no research governance index in laboratory of university, to compare with that proposed by the present research.

So, in discussion, first, fit measures are debated. Then, the methodology adopted in this work for the aggregation and standardization of the composite index is positioned with regard to the other methodologies used in this area.

The fit measures in this work are frequently used at the level of structural equations. However, they are also necessary when constructing a composite index (Muriithi et al., 2015). RMSEA, AVE, CR, and Cronbach’s alpha are among the most widely reported in the literature about fit measures related to composite index and to structural equations (Hair et al., 2019). Concerning these indicators, it was found that RMSEA is not very adequate when the sample size and degrees of freedom are small. However, small degrees of freedom do not tend to result in rejection of correctly specified models for the other indices (Taasoobsghiraz & Wang, 2016). However, the convergent validity is good when CR is higher than the AVE, and the AVE is higher than 0.5 (Hair et al., 2019).

In parallel, as the majority of fit indices depend on chi-square calculation, the chi-square test is autonomous (Cangur & Ercan, 2015) and constitutes a robust estimate when it is computed from the maximum likelihood extraction method (Lloret et al., 2017).

Taking into account the aforementioned findings and the items that this work incorporated, it is possible to assert that the indicators used for the elaboration of the composite index are well adjusted.

About the aggregation techniques, they take often additive or multiplicative forms with the use of several types of mean, such as arithmetic or geometric mean (Muriithi et al., 2015). For the standardization, the composite index frequently employs the usual normalization formulas, such as dividing all observations by the maximum of these observations or subtracting from each observation a pre-established minimum and dividing the result thus obtained by the difference between maximum and minimum pre-established. However, the problem of standardizing rating scales and indicators is sometimes posed as the case of Transparency International and Doing Business, which compute composite index by averaging the percentile ranks of countries on individual indicators. Consequently, to avoid drawbacks of habitual normalization methods and differentiation of notation scales, World Bank applies unobserved components methodology (Kaufman et al., 2010).

The elaborated method, in this work, is part of the methods that use linear programming, such as benefit-of-doubt model and synthetic indicator, proposed by Melyn and Moesen, with the exception that it performs aggregation and standardization of indicators simultaneously resulting in the standardized composite indicator. The elaborated standardized composite index is also different from indicators of the World Bank, which are based on perceptions rather than clear data (Apaza, 2008). In addition, these indicators do not respect the principle of discriminant validity (Langbein & Knack, 2010).

The method elaborated also differs from parametric methods, which necessitate several and sometimes complex tests, even if they are also based on synthetic scores, such as stochastic frontier analysis whose efficiency score, which synthesizes several variables values, is obtained by the ratio of the stochastic frontier production function to ordinary output function or on probabilities, such as discriminant analysis, whose scores are aggregated and standardized through the Bayesian formula (El Kadiri Boutchich, 2020a).

Conclusion

Conclusion includes the response to the problematic, implications, and limitations, their justifications and perspectives of this work.

Response to Research Problematic

An answer to the problematic raised in this work was given by insinuating that for the appropriateness of choosing, weighing, and aggregating indicators of the composite index, adequate methods are used. It is the question of Vigier Index and Adjusted DEA, then geometric mean, which are employed in two stages. Likewise, the validity and reliability of the elaborated indicator have been verified through adequate indicators.

Concerning Vigier index used for the adequate reconciliation of effectiveness and ethics, it constitutes an important theoretical and practical foundation for designing and implementing responsible and participatory research in higher education.

About Adjusted DEA, it is a type of DEA for which interesting improvements were envisaged, such as the use of qualitative variables (Sherman & Zhu, 2006) and negative or an interval data (Piri et al., 2016). Thus, Adjusted DEA is appropriate for composite index elaboration, as it deals with data of various types and gives optimal weights for indicators.

Moreover, the composite governance index developed could be applied in any context, provided that the number of variables is not very high, to avoid that important governance factors can be hidden in some components. The application of this index is facilitated by the availability of DEA software, such as DEAP, DEAOS, PIM-DEA, and Python for DEA.

Finally, the governance composite index developed can be considered as a composite index of ethics because governance and ethics are very interconnected, just as the indicators of participation, consultation, and transparency retained for the governance composite index elaboration, in this work, are frequently considered as ethical indicators (Dill, 2020).

Implications

The elaborated method has practical, research, and social implications. Regarding practical implications, the method developed being the result of the transformation of a method for measuring efficiency (DEA) is likely to improve both administrative performance through transparency and financial one (Kurdi, 2016). Financially, the combination of governance and efficiency ensures the autonomy of research laboratories (UNESCO, 2009) and makes it possible to properly manage research budgets in a transparent and efficient manner so as to have a balance between the growth rate of research laboratories and their operating expenses (Coccia, 2018). Similarly, this combination leads to the adoption of a societal management control and a renewed public management, which transcends New Public Management by associating trust (ethics) and performance (Pupion & Trébucq, 2018).

In addition to efficiency, this method conducts the supervisory ministry and the university to make decisions likely to improve governance, organization, and strategic management which are essential for the good performance of research laboratories (High Council for Evaluation of Research and Higher Education, 2015). Furthermore, as ethics’ practice is weak in laboratories, it is necessary to elaborate a code of ethics within them and expand it to the companies, so that these companies will be encouraged to invest in the research of the university’ laboratory and to finance it (van Wee, 2019). In fact, ethics’ practice has a main impact on innovation and performance of laboratory (Taebi et al., 2019).

Concerning research implications, this method encourages the integrated measures necessary to control the complexity of university activities (Perovic et al., 2017), as these measures are more informative and allow a better comparison in time and space (Bluszcz, 2016). The developed method also makes it possible to recalculate other fit measures, such as CFI, NFI, and Tucker-Lewis index from chi-square values to further improve the measurement of reliability and validity.

Finally, this method equally will have social implications if the research budget is allocated between laboratories according to their governance index scores because this index integrates ethics. This leads to the organizational justice that improves the motivation of research entities within the university and thus increases their scientific outputs (El Alfy, 2017).

Limitations, Their Justifications, and Perspectives

First, the sample size appears to be small. However, as already mentioned in the methodology, sample is composed by the most structured laboratories of the university, which are able to give information without any problem and without any bias. In addition, a small sample is acceptable when the factor analysis is linear and inter-indicators Pearson correlations are small (Lloret et al., 2017) as is the case of this study. Furthermore, the method used to aggregate the items is based on fractional programming (a linear program must be set for each entity). Consequently, unlike the parametric methods, which aggregate the results even if the sample is very large, the method used will pose results presentation problem, when the sample is very large. Thus, it will be very relevant notably for a comparison of groups of research entities, including laboratories, whose number is not very high.

Second, the variables incorporated in the composite index are not very numerous. However, this diminishes the main criticisms addressed to composite indicators, namely the fact of obscuring the source factors of governance, when a lot of variables are grouped (Apaza, 2008). In the same vein, composite indexes elaborated from great number of variables may not capture the interdependence of indicators, ignore important dimensions that are difficult to measure, and hide weaknesses in some components.

Third, to validity and reliability, other measures could be carried out to elaborate composite index, such as replicability, robustness, and accuracy of particular measures to underlying research questions (Gisselquist, 2014). However, validity and reliability constitute the main fit measures for relevance of new instrument (Taber, 2017).

Fourth, the method employed to determine the composite index gives relative results, with depend from benchmarks laboratories. However, it is possible to overcome this drawback using resulting meta-analysis indicators, such as the standardized mean difference (El Kadiri Boutchich, 2020d), to allow comparing the results, according to the proposed model, at local, regional, national, and international level.

The Vigier index calculation requires a sharp vision of the respondent to assign a score reconciling ethics and effectiveness. However, even if it could involve a dose of subjectivity, it is very useful for the normalization of the scores attributed to the ethics-effectiveness system and its weighing and its elasticity (the value of n) the laboratory wants to give to this system. In addition, the abovementioned subjectivity bias will not be exacerbated, as the respondent is the director of the laboratory whose experience will allow him to minimize this bias.

Finally, organizational and strategic management are presented in traditional way. Thus, it is interesting to utilize transversality, which allows the envisagement of a more efficient strategy and organization through adopting management by thematic or by homogeneous activities instead of management by teams (El Kadiri Boutchich, 2020d), which makes the discussion of the laboratory activity report and its strategic plan more fruitful.

As perspective, it is preferable to increase the number of variables in a convenient way. However, the perspective that seems the most important consists of combining parametric and non-parametric methods to determine the governance composite index (El Kadiri Boutchich, 2020e).

Footnotes

Authors’ Note

Driss El Kadiri Boutchich is a PhD in Economics; Graduate Diploma in Management Techniques (Data Analysis, Mathematical Programming, Inputs and Outputs methods, Statistics); and the Director of the Madeo laboratory.

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

ORCID iD

Driss El Kadiri Boutchich

References

Abdeldayem

M. M.

Aldulaimi

S. H.

(2018). Corporate governance practices in higher education institutions: The UK vs Bahrain. International Journal of Learning and Development, 8(4), 22–36.

Adam

Ovseiko

P. V.

Grant

Graham

K. E. A.

Boukhris

O. F.

Dowd

A.-M.

Balling

G. V.

Christensen

R. N.

, . . . International School on Research Impact Assessment. (2018). ISRIA statement: Ten-point guidelines for an effective process of research impact assessment. Health Research Policy and System, 16, Article 8.

Allen

M. L.

Garcio-Huidobro

Bastin

Hurtado

G. A.

Linares

Svetaz

M. V.

(2017). Reconciling research and community priorities in participatory trials: Application to Padres Informados/Jovenes Preparados. Family Practice, 34(3), 347–352.

Alzola

(2011). The reconciliation project: Separation and integration in business ethics research. Journal of Business Ethics, 99(1), 19–36.

Apaza

(2008). La pertinence des « indicateurs mondiaux de la gouvernance » : un débat scientifique en cours (The relevance of “global governance indicators”: scientific current debate). Revue Française D’administration Publique, 3(127), 591–599.

Arregui-Pabollet

Doussineau

Dettenhofer

(2018). An analytical framework to assess the governance of universities and their involvement in Smart specialisation strategies. Publications Office of the European Union. https://doi.org/10.2760/760453

Babcicky

(2013). Rethinking the foundations of sustainability measurement: The limitations of the Environmental Sustainability Index (ESI). Social Indicators Research, 113(1), 133–157.

Bayyurta

Yilmaz

(2012). The impacts of governance and education on agricultural efficiency: An international analysis. Procedia—Social and Behavioral Sciences, 58, 1158–1165.

Bluszcz

(2016). A comparative analysis of selected synthetic indicators of sustainability. Procedia—Social and Behavioral Sciences, 220, 40–50.

10.

Cai

Song

Xiao

Shi

(2020). The effect of social capital on Tacit knowledge-sharing intention: The Mediating role of employee vigor. SAGE Open, 10(3), 1–13. https://doi.org/10.1177/2158244020945722

11.

Cangur

Ercan

(2015). Comparison of model fit indices used in structural equation modeling under multivariate normality. Journal of Modern Applied Statistical Methods, 14(1), 152–167.

12.

Coccia

(2004). New models for measuring the R&D performance and identifying the productivity of public research institutes. R&D Management, 34(3), 267–280.

13.

Coccia

(2018). Metabolism of public research organizations: How do laboratories consume state subsidies? Public Organization Review, 19, 473–491.

14.

Council of Europe. (2012). Guide for research ethics committee members. https://www.coe.int/en/web/bioethics/guide-for-research-ethics-committees-members

15.

Ddungu

Edopu

(2016). Social responsibility of public and private universities in Uganda. Makerere Journal of Higher Education, 8(1), 71–90.

16.

Dill

D. D.

(2020). Enhancing academic quality and collegial control: Insights from US policy on the ethical conduct of Human subjects’ research. Higher Education Policy, 33, 45–64.

17.

DuBois

J. M.

Antes

A. L.

(2018). Five dimensions of research ethics: A stakeholder framework for creating a climate of research integrity. Academic Medicine: Journal of the Association of American Medical Colleges, 93(4), 550–555.

18.

El Alfy

. (2017). Investigating organisational justice in higher education in UAE. International Journal of Management in Education, 11(2), 163–187.

19.

El Gibari

Gómez

Ruiz

. (2019). Building composite indicators using multicriteria methods: A review. Journal of Business Economics, 89, 1–24.

20.

El Kadiri Boutchich

. (2019). Proposal of an alternative human capital evaluation method for a research laboratory: Multidimensional approach based on physical indicators and activities, Journal of Education, 199(1), 35–44.

21.

El Kadiri Boutchich

. (2020a). Efficiency in the Public University: Towards an Adequate Measuring of Research Laboratories. International Journal of Applied Management Sciences and Engineering, 7(1), 100–121.

22.

El Kadiri Boutchich

. (2020b). Human capital in public research laboratories: A method proposal for assessing actual and target scores. Journal of Applied Research in Higher Education, 12(5), 1189–1208.

23.

El Kadiri Boutchich

. (2020c). Factors with significant impact on efficiency of research laboratories: Case of the public university. Quality & Quantity, 54(4), 1317–1333.

24.

El Kadiri Boutchich

. (2020d). Human capital in public research laboratory: Towards an alternative evaluation and prediction method based on hybridization. Annals of Data Science, https://doi.org/10.1007/s40745-020-00241-9

25.

El Kadiri Boutchich

. (2020e). Relevant ingredients for identifying factors with significant impact on research structures efficiency in higher education. Journal of Education. https://doi.org/10.1177/0022057420914910

26.

Ferretti

Pereira

A. G.

Vértesy

Hardeman

(2018). Research excellence indicators: Time to reimagine the ‘making of’? Science and Public Policy, 45(5), 731–741.

27.

Ghewy

(2010). Guide pratique de l’analyse de données, avec applications sous IBM SPSS STATISTICS et Excel, questionnez, analysez. . . et décidez [A practical guide to data analysis with SPSS and Excel software, investigate, analyze. . . and decide]. De Boeck.

28.

Gisselquist

R. M.

(2014). Developing and evaluating governance indexes: 10 questions. Policy Studies, 35(5), 513–531.

29.

Hair

J. F.

Black

W. C.

Babin

B. J.

Anderson

R. E.

(2019). Multivariate data analysis (8th ed.). Cengage.

30.

High Council for Evaluation of Research and Higher Education. (2015). IFSTTAR evaluation report. https://www.ifsttar.fr/fileadmin/redaction/2_recherche-expertise/2.7_evaluation/B2017-EV-0772765M-S1S1170013138-016543-RD.pdf

31.

High Council for Evaluation of Research and Higher Education. (2020). Evaluation report on the research unit: Institute of human genetics. https://www.igh.cnrs.fr/images/equipes/seitz/HCERES_Final_Report_Team20Seitz.pdf

32.

Kao

Liu

S. T.

(2011). Efficiencies of two-stage systems with fuzzy data. Fuzzy Sets and Systems, 176, 20–35.

33.

Kaufman

Kraay

Mastruzzi

(2007). Growth and governance: A reply. The Journal of Politics, 69(2), 555–562.

34.

Kaufman

Kraay

Mastruzzi

(2010). The Worldwide governance indicators. Methodology and analytical issues (Policy Research Working Paper 5430). The World Bank. https://openknowledge.worldbank.org/bitstream/handle/10986/3913/WPS5430.pdf?sequence=1&isAllowed=y

35.

Khiari

Karaa

Omri

(2007). Corporate governance efficiency: An indexing approach using the stochastic frontier analysis. Corporate Governance, 7(2), 148–161. https://doi.org/10.1108/14720700710739796

36.

Kurdi

(2016). Good governance and its impact in improving administrative and financial performance: A case study of Nablus municipality. International Humanities Studies, 3(4), 30–47.

37.

Kwiek

(2015). The unfading power of collegiality? University governance in Poland in a European comparative and quantitative perspective. International Journal of Educational Development, 43, 77–87.

38.

Langbein

Knack

(2010). The worldwide governance indicators: Six, one, or none? Journal of Development Studies, 46(2), 350–370.

39.

Lloret

Ferreres

Hernández

Tomás

(2017). The exploratory factor analysis of items: Guided analysis based on empirical data and software. Anales de Psicología, 33(2), 417–432.

40.

Martineau

(2017). De quoi les outils de gestion sont-ils faits ? La structure « listique » des artefacts de gestion [What are the management tools made of? The “list” structure of management artefacts]. M@n@gement, 20(3), 239–262.

41.

McCracken

(2020). Ethics as obligation: Reconciling diverging research practices with marginalized communities. International Journal of Qualitative Methods, 19, 1–11.

42.

Melyn

Moesen

(1991). Towards a synthetic indicator of macroeconomic performance: Unequal weighting when limited information is available (Public Economics Research Paper, 17, CES, KULeuven).

43.

Muriithi

Jimenez

Jannin

Sajid

Singh

Sharma

(2015). Quantifying governance: An indicator-based approach. London School of Economics/Department of International Development. https://www.gov.uk/research-for-development-outputs/quantifying-governance-an-indicator-based-approach

44.

Oancea

(2019). Research governance and the future(s) of research assessment. Palgrave Communications, 5, Article 27.

45.

OECD (2008). Handbook on constructing composite indicators. Methodology and user guide. https://www.oecd.org/sdd/42495745.pdf

46.

Perovic

Bojanic

Nerandzic

(2017). Measuring efficiency of teaching process and Faculty in Transition States Using DEA analysis. Technical Gazette, 24(5), 1509–1517.

47.

Piri

Givehchi

S. H.

Keshavarz

(2016). Computing efficiency for decision making units with negative and Interval data. Journal of Data Envelopment Analysis and Decision Science, 1, 5–14.

48.

Pupion

P.-C.

Trébucq

(2018). Un management public en recherche de confiance, équilibre et performance [Editorial; A public management in quest for confidence, balance and performance]. Gestion et Management Public, 7(1), 6–8.

49.

Ribes

Baker

(2007). Modes of social science engagement in community infrastructure design. In Steinfield

Pentland

B. T.

Ackerman

Contractor

(Eds.), Communities and technologies. Springer.

50.

Salimi

Rezaei

(2016). Measuring efficiency of university-industry Ph.D. projects using best worst method. Scientometrics, 109(3), 1911–1938.

51.

Savall

Zardet

(2011). Le modèle de gestion socioéconomique dans les organisations de service public. Applications au Mexique, en Belgique et en France [The socioeconomic management model in public service organizations. Cases of Mexico, Belgium and France]. Réunion annuelle OCDE—Paris—7 novembre. http://www.oecd.org/gov/pem/49145143.pdf

52.

Shaw

Boynton

et Greenhalgh

(2005). Research governance: Where did it come from, what does it mean? Journal of the Royal Society of Medicine, 98(11), 496–502.

53.

Sherman

H. D.

Zhu

(2006). Benchmarking with Quality-Adjusted DEA (Q-DEA) to seek lower-cost high-quality service: Evidence from a U.S. bank application. Annals of Operations Research, 145(1), 301–319.

54.

Streimikiene

Baležentis

Kriščiukaitienė

(2016). Benefit of the doubt model for financial risk analysis of Lithuanian family farms. Economics and Sociology, 9(1), 60–68.

55.

Sugarman

(2016). Ethics and regulatory challenges and opportunities in patient-centered comparative effectiveness research. Academic Medicine, 91(4), 455–457. https://doi.org/10.1097/ACM.0000000000001096

56.

Taasoobsghiraz

Wang

(2016). The performance of SRMR, RMSEA, CFI, and TLI: An examination of sample size, path size, and degrees of freedom. Journal of Applied Quantitative Methods, 11(3), 31–39.

57.

Taber

K. S.

(2017). The use of Cronbach’s alpha when developing and reporting research instruments in science education. Research in Science Education, 48, 1273–1296.

58.

Taebi

van den Hoven

Bird

S. J.

(2019). The importance of ethics in modern universities of technology. Science and Engineering Ethics, 25, 1625–1623.

59.

UNESCO. (2009). The new dynamics of higher education and research for societal change and development. https://www.inqaahe.org/sites/default/files/UNESCO%20communique.pdf

60.

University of Edinburgh. (2020). REF and research policy. https://www.ed.ac.uk/governance-strategic-planning/research

61.

van Wee

. (2019). The need for a code of conduct for research funders. Commentary on “Values in university-industry collaborations: The case of academics working at universities of technology.” Science and Engineering Ethics, 25(6), 1657–1660.

62.

Vertesy

(2018). The Adjusted Research Excellence Index 2018. Methodology Report, EUR 29567 EN, Publications Office of the European Union, Luxembourg, ISBN 978-92-79-98383-2, https://doi.org/10.2760/126300, JRC114345

63.

Welch

(2020). Tensions in the evolving Australian higher education system: A complex, evolving mix. Higher Education Governance and Policy, 1(1), 32–48.

64.

Wight

J. B.

(2017). The ethics behind efficiency. The Journal of Economic Education, 48(1), 15–26.

65.

Ysa

Albareda

Forberger

(2014). What is governance? In Peter

Gerhard

Toni

(Eds.), Reframing addictions: Policies, processes and pressures (Chapter 2, pp. 6–16). ALICE-RAP.