Disintegrating the effect of market orientation on firm performance
A meta-analytic comparison across four geographic regions
Master Thesis in International Business and Management Author: Lynn Overweg
Student number: S3731960 Supervisor: Dr. C. Schlägel Co-assessor: Dr. J. Shin
Date of submission: 14-05-2020 Word count: 13,532
University of Groningen
Faculty of Economics and Business
1
Abstract... 2
1. Introduction ... 3
2. Theoretical background ... 5
2.1 Dimensionality and conceptualization of market orientation ... 5
2.2 Towards a mutualism perspective of MO dimensions ... 7
2.3 Country specific effects of market orientation ... 11
2.4 The moderating role of contextual and methodological factors ... 11
3. Research methodology ... 14
3.1 Literature search ... 14
3.2 Sample and coding ... 14
3.3 Data procedure... 15
4. Results ... 17
4.1 Results of bivariate meta-analysis ... 17
4.2 Results of the commonality analysis ... 18
4.3 Results of the moderator analysis ... 22
5. Discussion... 34
5.1 Theoretical implications ... 34
5.2 Practical implications ... 36
5.3 Limitations and directions for future research ... 38
References ... 40
Appendix ... 46
2 Abstract
The importance of taking on a market-oriented approach has grown significantly over the years.
This has led to an increasing number of company’s which now focuses on generating a thorough understanding of the market and its customers´ needs and preferences, which allows it to create superior customer value. How this concept exactly affects a firms’ performance has already been studied in various ways and for different regions, with mostly positive results. However, a study which disintegrates the common, bilateral, and unique effects of the three dimensions of which market orientation exists, is lacking. This study will address this literature gap by combining the resource-based view with a mutualism perspective and commonality analysis to observe these three different levels of analysis and compare the results across four different regions: Europe, Asia, the U.S., and Africa. By performing a meta-analysis, which combines a total of 44 studies on the effect of market orientation on firm performance, the effect size of this research will be improved. The study will thus add to the existing literature by providing more precise effect size estimates and a more detailed and comprehensive theoretical understanding of the mechanisms through which market orientation influences firm performance.
Keywords: Market orientation; meta-analysis; commonality analysis; firm performance
3 1. Introduction
Over the course of time, the world has changed from a place where most people were only concerned with producing enough food and clothing in order to survive, with barely any alternatives to choose from, to a place where there is more choice than ever before (Szmigin and Piacentini, 2018). Even geographical boundaries which used to limit consumer’s options barely seem to play a role anymore (Khandare, 2011). Companies thus have to put in a great deal of effort to outperform the competition and persuade customers to choose their products over those of its competitors (Szmigin & Piacentini, 2018). It is therefore not surprising that many companies take on a market-oriented approach, in which they focus on generating a thorough understanding of their customers’ current and future needs and preferences and share these findings across the organization, to create superior customer value (Kohli & Jaworski, 1990; Slater & Narver, 1990), whilst at the same time meeting the needs of other stakeholders (Langerak, 2003). Market oriented companies actively gather knowledge about their customers and competitors and aim to effectively integrate this information in their operations (Hansen, Dibrell & Down, 2006). This knowledge allows them to create market differentiated products (Vieira, 2010) which are of better quality, more innovative or of higher value as compared to the offering of their competitors (Ge & Ding, 2005). As a result, companies are able to generate a sustainable competitive advantage which will eventually lead to higher organizational performance (Hunt & Morgan, 1995).
Ever since the introduction of MO in the 1990s, it has been a popular topic of research and many of the MO literature has been dedicated to determining if and how MO actually improves firm performance. Throughout these nearly thirty years of research, even though some studies found a non-significant or mixed relation between MO and firm performance (e.g., Shoham &
Rose, 2001), the majority of studies and multiple meta-analyses have confirmed a positive relation between MO and various measures of firm performance (e.g., Baker & Sinkula, 1999;
Ellis, 2006; Grinstein, 2007; Homburg & Pflesser, 2000; McNaughton, 2001; Pelham &
Wilson, 1996; Ruekert, 1992; Shoham, Rose & Kropp, 2005; Vieira, 2010). Over time, two main perspectives on MO have emerged (Homburg & Pflesser, 2000). The first perspective, that of Jaworski and Kohli, is behavioral in nature and believes that customer value can be created by generating and responding to market intelligence (Han, Kim & Srivastava, 1998;
Spillan, Kara, King & McGinnis, 2013). According to this view, the components of MO thus consist of intelligence generation, intelligence dissemination, and responsiveness (Jaworski &
Kohli, 1993). The second perspective, that of Narver and Slater, approaches MO in terms of
4 culture and beliefs that it is related to the fundamental characteristics of the organization (Cano, Carrillat & Jaramillo, 2004; Homburg & Pflesser, 2000). This cultural perspective considers the components of MO to be customer orientation, competitor orientation, and interfuctional coordination (Narver & Slater, 1990). For the sake of this research, the three MO components of Narver and Slater and the three MO elements of Jaworski and Kohli will be called dimensions.
When measuring the effect of MO on firm performance, the approaches of researchers have not always been consistent. In particular, there seems to be no clear agreement among researchers whether MO should be measured as an overall construct (e.g., Matear, Osborne, Garrett & Gray, 2002; Olavarrieta & Friedmann, 2008; Zhou, Li, Zhou & Su, 2008), or whether the focus should be on the separate effects of the dimensions of either Jaworski and Kohli or Narver and Slater (e.g., Han et al., 1998; Im & Workman, 2004; Lukas & Ferrel, 2000; Noble, Sinha & Kumar, 2002). Criticism on both approaches can be thought of. When measuring MO as an overall construct, for instance, researchers are only aware of the size of the relation between MO and performance, they know nothing about which dimension or combination of dimensions is contributing most. As a result, researchers and practitioners do not know whether it is more important for a certain organization to invest more money in a particular dimension rather than investing an equal amount in all three dimensions. In contrast, when studies only focus on the unique effect of the individual dimensions, they face the risk of misinterpreting certain information. To illustrate, if it turns out that two of the dimensions have a significant effect on performance and one of them does not, researchers may wrongly assume that the third dimension is therefore not important. Because even though on its own this dimension may not seem relevant, it may be the case that the dimension still greatly contributes to the common effect of MO on performance by reinforcing the other dimensions. And when this dimension is neglected, it may be harmful to the effect of the overall construct on performance.
Thus, despite the considerable amount of research done on the effect of MO on firm
performance, meta-analyses thus far have mainly focused on the effect of the overall construct
of MO, or on the individual dimensions of either Jaworski and Kohli, or Narver and Slater. A
limited number of studies focuses on both. In addition, there is no knowledge of a meta-analysis
which examines the bilateral performance effects of any combination of only two dimensions
(i.e., how one of the dimensions influences another in a bi-directional way). By drawing on the
resource-based view and combining it with a mutualism perspective, this study is therefore
aiming to fill this research gap to provide a clear understanding of the various ways in which
5 MO affects firm performance. In doing so, this study will contribute to the existing body of literature in four ways. First, by performing a meta-analysis, this study provides more precise effect size estimates and further evidence on the influence of MO on firm performance. Second, through the inclusion of a commonality analysis, this study allows to distinguish between the effect of MO on performance on three different levels; the effect of the overall MO construct, the unique effect of the individual dimensions, and the shared effect of the combination of any two dimensions. Third, this study will distinguish between the effects of the two major MO scales, which allows to compare the two in terms of similarity. Finally, this study will compare the results of the meta-analysis across four geographic regions: Europe, Asia, the U.S. and Africa. This study will thereby determine if and to what extent the effects or combination of effects will be stronger for certain regions (e.g., whether in some region customer orientation has a stronger effect than competitor orientation). In the end, this study will provide both theoretical and practical implications based on the results of the meta-analysis. In line with the above reasoning, the research question to be addressed in this paper is: To what extent do the unique, bilateral and overall effect of market orientation affect firm performance and to what extent do the results differ across Europe, Asia, the U.S., and Africa?.
The structure of the remainder of this paper is as follows: the second chapter contains the theoretical background of the study, the third chapter describes the methodology behind the data analysis, and the fourth chapter presents the results. Finally, the fifth chapter entails the discussion, which includes the theoretical implications, practical implications, and limitations, which will be translated in directions for future research.
2. Theoretical background 2.1 Dimensionality and conceptualization of market orientation
Due to the excessive set of product and service choices consumers nowadays have, businesses
increasingly want to understand why people consume and shop the way they do, in order to
create successful products and communicate effectively (Szmigin & Piacentini, 2018). As a
result, an increasing number of companies is now taking on a market-oriented approach, which
can be defined as ‘the organizational culture that most effectively and efficiently creates the
necessary behaviors for the creation of superior value for buyers and, thus, continuous superior
performance for the business’ (Narver & Slater, 2000: 21). Since its introduction in the 1990s
many researchers have attempted to find out whether MO actually has an effect on firm
performance. And, throughout the years, several meta- analyses were able to confirm the
6 positive relation between MO and various measures of firm performance (e.g., Chang, Franke, Bulter, Musgrove & Ellinger, 2014; Shoman et al., 2005).
This positive relation between MO and firm performance can be reasoned back to the resource- based view (i.e., the theoretical view that a firm’s internal resources can create a sustainable competitive advantage), introduced by Barney (1991). Market oriented companies actively gather knowledge about their customers and competitors and integrate this information in their operations (Hansen et al., 2006). Day (1994), suggests that market-oriented firms can use their knowledge to create certain capabilities which allows those firms to generate a sustainable competitive advantage. And due to these capabilities, firms will be able to offer higher quality or more innovative products and services which allows the firm to set higher prices than its competitors (Ge & Ding, 2005). Thus, being market oriented provides a source of competitive advantage to a firm (Morgan, Vorhies & Mason, 2009). In a similar vein, Baker and Sinkula (1999) propose that firms which are strongly market oriented and able to facilitate discontinuous innovation are more likely to create a sustainable competitive advantage. This sustainable competitive advantage will in turn result in sustained superior performance of the firm (Barney, 1991; Hunt & morgan, 1995).
In the existing literature, there are two dominant scales to measure MO. The first scale, called MKTOR, is developed by Narver and Slater (1990). This scale originally consisted of 21 items which together measure three behavioral components of an organization and reflect the underlying norms and values of that organization (Homburg & Pflesser, 2000; Kirca, Bearden, Tomas & Hult, 2011). According to Narver and Slater (1990) the three behavioral components are: customer orientation (i.e., gaining an understanding of customer demands), competitor orientation (i.e., gaining an understanding of how the firm’s competitors are satisfying these demands), and interfunctional coordination (i.e., cooperation and the flow of knowledge across the organization). The second scale, developed by Jaworski and Kohli (1993), is called MARKOR. This scale originally consisted of 20 items to measure its three elements;
intelligence generation (i.e., gaining of an understanding of the needs and preferences of
customers and the factors which influence those needs and preferences), intelligence
dissemination (i.e., the spread of market intelligence to relevant departments and employees
within the organization), and responsiveness (i.e., responding to generated and disseminated
needs and preferences of the market). The three components of MO of Narver and Slater and
three elements of MO of Jaworski and Kohli, which are called dimensions in this research, are
visualized in figure 1a and b below.
7 Figure 1: Conceptualizations of MO
1a. Dimensions of Narver and Slater (1990) 1b. Dimensions of Jaworski and Kohli (1993)
2.2 Towards a mutualism perspective of MO dimensions
As mentioned before, there is a lot of inconsistency in the measures and findings in the existent literature on the relation between MO and firm performance. Several meta-analyses have aimed to provide clarity by synthesizing the empirical results of the individual studies (e.g., Grinstein, 2007). Table 1 below summarizes the objectives and main findings of these meta-analyses.
When analyzing this existent literature it becomes clear that the majority of the studies focus only on the effect of the overall construct of MO on firm performance, a significantly lower amount focuses on the effect of the individual dimensions on performance and an even more limited number of studies focuses on both. In particular, performance effects of any combination of only two of its dimensions are still unexplored (e.g., the combined effect of customer orientation and competitor orientation or intelligence generation and intelligence dissemination on performance). When studies do focus on the effect of the individual dimensions, most of them treat them as separate and independent constructs which affect performance in parallel, as visualized in figure 2a (e.g., Ansah & Chinomona, 2017; Chao &
Spillan, 2010; Chin, Lo & Ramayah, 2013). A few studies do suggest a level of interdependence among the dimensions and propose a mediator effect, as visualized in figure 2b and 2c.
Carbonell and Rodríguez Escudero (2010), for example, proposed the responsiveness as a
mediator between the impact of intelligence generation and intelligence dissemination on
performance. Furthermore, Grinstein (2007) established that the effect of competitor orientation
depends on a minimum level of customer orientation. In addition, Im and Workman (2004)
concluded that the effect of customer orientation and competitor orientation on performance is
facilitated by interfunctional coordination. Nevertheless, these dimension interdependencies are
still under-researched and a meta-analysis which specifically focuses on this topic is lacking.
8 Table 1: Meta-analytic findings
Authors MO scale Performance type Study objective Key findings
Rodríguez Cano et al.
(2004)
Both Business performance
Identify the strength of the MO-performance relationship and the impact of various moderators
• Positive relation between MO and business performance
• Stronger relation for service firms and MARKOR scale
Shoham et al. (2005) Both Organizational performance
Identify the robustness of the MO-performance relationship under various conditions.
• Positive relation between MO and performance
• No significant difference between the MO scales
Kirca et al. (2005) Both Organizational performance
Identify the antecedents of MO and the way in which MO influences performance
• MO-performance relation is weaker for service firms and stronger in low power-distant cultures
Ellis (2006) Both Business and
financial performance
Identify whether the MO-performance relation differs across country’s measures or contexts
• Each region showed positive correlations between MO and performance, this effect was the strongest in the U.S.
• Higher effect sizes for the MARKOR scale
Grinstein (2007) MKTOR Innovation performance
Identify the effect of MO dimensions on innovation performance and the impact of multiple contextual characteristics
• All MO components positively and equally affect innovation performance
• The effect of competitor orientation depends on a minimum level of customer orientation
Vieira (2007) Both Business
performance
Identify the impact of MO on performance in Brazil and international studies and the role of several moderators
• Positive and significant association between MO and business performance for both studies
• Stronger correlations for the MARKOR scale
Chang et al. (2014) MKTOR NPP and firm performance
Identify whether radical and incremental innovation mediate the MO-performance relationship and test if these results differ between goods and services
• Both types of innovation positively effect performance
• MO positively affects firm performance
9 Figure 2: MO dimension dependencies
2a. Independent MO dimensions 2b. Dependent MO dimensions
2c. Mediational chain of MO dimensions
Both the independent and dependent views on the dimensions have been proven to be true in the existing literature. Nevertheless, there is no knowledge of a study or meta-analysis which combines both views to gain clarity on the multidimensional effect of MO on firm performance.
Building on the mutualism perspective (Van Der Maas, Dolan, Grasman, Wicherts, Huizenga
& Raijmakers, 2006) this study is therefore aiming to combine both views to identify that the change of a firm’s performance can both be attributed to the independent effect of the individual dimensions, as well as to the interrelation of at least two of the dimensions. In a similar vein, Lomberg et al. (2016) have looked in their study at entrepreneurial orientation as both a unidimensional and multidimensional construct. Thus, besides analyzing the overall effect of entrepreneurial orientation on firm performance, the authors have also looked at the individual and bilaterally shared effects of its dimensions. This study is aiming to do the same for market orientation.
For the dimensions of Narver and Slater (1990) for instance, firm performance may increase
even more when the information on the needs and preferences of customers (i.e., customer
orientation), is shared across the other departments of the organization (i.e., interfunctional
coordination). If, for example, the marketing department shares newly gained knowledge on
the preferences of its customers with the R&D and sales departments, the R&D department can
use this knowledge to create products or services which will be of interest to the customers and
the sales agents will know which features of the product or service to emphasize when trying
to sell it. In addition, the interaction between these dimensions may create additional mutual
10 benefits for both. To illustrate, the sales or R&D departments may identify some gaps in the information provided by the marketing department and therefore request follow up information which may provide new insights which may benefit the entire organization. A similar logic can be applied to the competitor orientation and interfunctional coordination and to the intelligence generation, intelligence dissemination and responsiveness dimensions of Jaworski and Kohli (1993).
This paper thus aims to combine the RBV with a mutualism perspective, to study the extent to which the effect of MO on firm performance is explained by the effect of the individual dimensions, the combination of any of the two dimensions (i.e., bilateral effect), or the effect of the overall construct. This reasoning is visualized in figure 3. The individual dimensions of Narver and Slater (1990) are represented as Ecu (i.e., individual effect of customer orientation), Ein (i.e., individual effect of interfunctional coordination) and Eco (i.e., individual effect of competitor coordination). The bilaterally shared effect of customer orientation and competitor coordination is represented as Ecu,co. that of competitor orientation and interfunctional coordination Eco,in, and that of customer orientation and interfunctional coordination Ecu,in.
Their overall shared effect is labeled as Eo. The same logic can be applied to the dimensions of Jaworski and Kohli (1993). Intelligence generation is labeled Ege, intelligence dissemination as Edis, and responsiveness as Eres. Their bilaterally shared effects are labeled as Ege,di, Edi,re, and Ege,re. Their overall effect is labeled Eo as well. The six dimensions together represent V (i.e., the variance of firm performance that is explained by MO).
Figure 3: The variance of firm performance explained by MO
Note: Visualization adapted from Lomberg et al. (2016)
11 Based on the above reasoning, this leads to the following hypothesis:
Hypothesis 1: The effect of MO on firm performance can be explained by the common effect of the dimensions, as well as the unique and bilateral effects.
2.3 Country specific effects of market orientation
Several studies on the linkage between MO and performance have established that the strength of this linkage is dependent on contextual moderators. Previous studies determined that a firm is not only affected by external factors (Ellis, 2010), its outcome is also determined by the firm’s marketplace (Ellis, 2006). In addition, the meta-analysis of Vieira (2007) established that gaining market oriented knowledge is more important in country’s with a stronger competitive environment, as it is an important tool for firms to innovate their product offerings and gain an advantage over their competitors (Vieira, 2007). Furthermore, this author also concluded that collecting market information is less important in countries with highly technological turbulent environments, as the creation of new products is mainly driven by R&D efforts, rather than the needs of customers. In his attempt to discover whether the relation between MO and performance is universal across different countries, the meta-analysis of Ellis (2006) already established that effect sizes differed depending on cultural and economic characteristics of a country, with effect sizes being significantly the strongest in the USA. This implies that focusing on MO will have varying effects on the performance of organizations depending on the region they are in. However, as the previously mentioned study only focuses on the overall MO construct, there is no knowledge of a study which compares the effect sizes of the individual dimensions across regions (e.g., whether in Europe customer orientation has a higher effect on firm performance than in Asia, or if intelligence generation in the U.S. has a higher effect on firm performance than in Europe or the other way around). Therefore, the question of whether this is also the case for the individual MO dimensions remains unanswered. As previous literature established that the effect sizes of the overall construct differ across regions, it is also expected that the results will vary across the individual dimensions. Hence:
Hypothesis 2. The effects of the dimensions of MO on firm performance will differ for the four different regions: Europe, Asia, the U.S., and Africa.
2.4 The moderating role of contextual and methodological factors
To determine whether the proposed relationship between market orientation and firm
performance is influenced by potential moderators, several methodological and contextual
12 moderators were coded. The contextual moderators include industry (i.e., service, manufacturing, or mixed industries) and culture (i.e., high vs. low power distance) and the methodological moderators include performance measure (i.e., financial or organizational performance) and year of data collection. Figure 4 presents a visual representation of the different variables which will be tested as potential moderators of the conceptual model. The reasoning behind the moderators will be explained in more detail below.
Figure 4: Potential moderator variables
Contextual moderators
The first contextual factor which will be included in the moderator analysis is firm industry type, i.e., manufacturing, service or mixed industries. The original MO study of Narver and Slater (1990) found that the relation between MO and performance is stronger for firms in service industries than for firms in manufacturing industries. This is in line with the paper of Avlonitis and Gounaris (1997) which explains that manufacturing firms tend to be more sales oriented over market oriented as compared to service firms. Nevertheless, the latter study also shows that the relationship between MO and company performance for industrial firms is positively and significantly associated with all the performance measures they included, whereas for service firms this was only the case for some of the measures (Avlonitis &
Gounaris, 1997). Furthermore, even previous meta-analyses, where the results of various
studies are combined, are still inconsistent in their findings. The study of Kirca et al. (2005),
for instance, concludes that the relationship between MO and performance for firms in the
service industry is weaker than for firms in the manufacturing industry. In contrast, the meta-
analysis of Rodríguez Cano et al. (2004) shows that the MO performance relationship is
stronger for service than for manufacturing firms. This presents the opportunity to use this study
13 to shed light on these contradictive results, by comparing the results across the different dimensions. Hence, industry type is included as a moderator in this analysis.
The second potential moderator of the relation between MO and performance is culture, more specifically high versus low power distance. Power distance is known as the extent to which subordinates accept and expect unequal power distribution in a company (Hofstede, 2001).
Following the reasoning of the meta-analysis on entrepreneurial orientation of Saeed et al.
(2014), in countries where there is a high degree of power distance between superiors and subordinates, subordinates are less inclined to present their own innovative ideas. As a result, there is a lower rate of innovative ideas and products in these countries. In turn, this may negatively affect the motivation of subordinates to generate and disseminate new information which could lead to products which are in line with the preferences of customers or ahead of the products of competitors. In addition, it might also hamper the coordination between different departments and functions and ultimately thus negatively affect performance. To test this reasoning, power distance will be included as a potential moderator in this analysis.
Methodological moderators
The first methodological factor to be included in the moderator-analysis is performance measure. A large number of studies measure performance either as financial performance (e.g., Balas, Gokus & Colakoglu, 2014; Boso, Story & Cadogan, 2013; Morgan et al., 2009) or business/organizational performance (e.g., Brik, Rettab & Mellahi, 2010; Langerak, 2003; Sin, Tse, Yau, Lee, Chow & Lau, 2000). This allows for the opportunity to test whether the effect size between MO and performance is stronger in terms of financial or organizational performance, or whether this makes no difference. Hence, performance measure will be included as a potential moderator.
The second and final methodological factor which will be covered in the moderator-analysis is
year of data collection. Several studies have proven the importance of MO by showing a
positive relation between a firm’s MO and performance. This meta-analysis allows for these
results to be compared to see whether the effect of MO on performance for more recent studies
is bigger than for older studies (and the importance of MO thus has increased over the years),
or whether it is the other way around. In light of this, year of data collection will be incorporated
as a moderator in this analysis. Again, results will be shown for each individual dimension.
14 3. Research methodology
3.1 Literature search
The articles included in the meta-analysis were found in three ways. First, through EBSCO and Google Scholar by typing in relevant keywords and combinations of keywords (e.g., effect market orientation on performance and effect market orientation dimensions on business performance). Second, by scanning the articles included in other meta-analyses on the relation between MO and firm performance (e.g., Ellis, 2006; Shoham et al., 2005). Lastly, by looking at the reference list of various individual papers on the effect of MO on performance which referred to other studies performing similar analyses.
Despite the considerable amount of papers which study the relationship between MO and performance, the majority could not be included in this meta-analysis. For a paper to be considered relevant for this study, it had to meet a few requirements. First of all, the paper had to include the correlation coefficients of the individual dimensions of MO and performance or provide information in such a way that it was possible to calculate the correlations. Papers that only focused on the overall construct of MO were not relevant for this study. Second of all, studies were excluded if they only focused on one or two of the three dimensions of the two measurement scales. Third, the dependent variable; firm performance, should either be measured as financial performance or organizational/business performance. Studies on other types of performance, e.g., export or innovation performance were excluded. In addition, only papers that measure MO by using either the MKTOR or MARKOR scale or both were relevant to include. Furthermore, only studies which were based in countries of one of the four regions were included, as for the other regions there was not enough literature available to be included in the meta-analysis. And finally, only English language studies were included in the sample.
3.2 Sample and coding
A total number of 42 studies with unique samples were coded and included in the meta-analysis,
of which two studies included two separate samples and results for two different countries (i.e.,
Chao and Spillan (2010) and Jangl (2015)). Of these 42 studies, 22 measured MO using the
MKTOR scale and 18 using the MARKOR scale. Two studies included both the MARKOR
and MKTOR scale. The following information for all these studies was coded; sample size,
year of data collection, country culture score (high vs. low power distance), firm industry type,
MO measurement scale, reliability of MO dimensions and performance (Cronbach’s alpha),
item to item correlation, item to performance correlation and overall MO to performance
15 correlation. Culture was coded using the Hofstede country comparison data (Hofstede 2020).
Additionally, performance was coded separately for organizational performance and financial performance. The year of data collection of the studies ranges from 1986 to 2016. When the year of data collection was not mentioned in the study, the year of submission minus two years is used instead. If the year of submission was not mentioned either, the year of publication minus three years was used. Furthermore, if the Cronbach’s alpha of a study was not mentioned, the mean Cronbach’s alpha of the other studies combined was used instead. To prevent errors being made in the coding of the studies, each study was coded twice in a separate scheme and then compared to see if the two schemes were identical. The number of firms included in the samples ranges from 29 to 451. The total sample size includes 9,185 firms. The geographic dispersion of the studies is as follows: thirteen studies were performed on firms from European countries, four African countries, thirteen Asian countries, nine American countries and five studies included firms from multiple countries in their sample. In this study, even though both countries are located in both Europe and Asia, based on the geographic location of the capitals of the countries, Russia is classified as a European country and Turkey as an Asian country.
Furthermore, thirteen studies measured the relation between MO and performance in service industries, fourteen in manufacturing industries and the fifteen in mixed industries (i.e., a variety of industries other than manufacturing or service). Finally, 26 studies measured performance as financial performance (e.g., return on assets or return on equity), fifteen as organizational performance (e.g., reputation and customer retention rate), and one measured both. An extensive overview of the main information of the coded papers included in the meta- analysis can be found in table A1 in the appendix.
3.3 Data procedure
This study follows the data procedure of Hunter and Schmidt (2004) and that of other MO meta-
analyses (e.g., Ellis, 2006). First, the dependent and independent variables of the studies
included in the meta-analysis were corrected for measurement error. This is recommended
because the methods used to measure these variables varies across the different studies. As a
result, these measures differ in the amount to which they are affected by measurement error,
which may lead to false conclusions (Hunter & Schmidt, 2004). To correct for measurement
error, the effect sizes are adjusted for their reported reliability (i.e., Cronbach’s Alpha). If
studies did not report their reliability, the mean construct reliability was used instead. Second,
the findings were corrected for sampling error. This is recommended because studies with
larger sample sizes have a higher probability to obtain a more precise estimation than studies
16 for which the sample size is smaller (Ellis, 2006). To correct for sampling error, the estimation of the mean effect size is weighed against the different sample sizes of the included studies.
Finally, the amount of heterogeneity of the effect size is assessed by calculating the Q-score and I² value, both of which are reported in the results tables.
3.4 Meta-analysis overview
To gain a clear and common understanding of all the important variables and moderators included in this analysis, table 2 presents an overview of all the constructs and their definition.
Table 2: Overview of variables and moderators
Variables: Explanation:
Customer orientation Gaining an understanding of customer demands (Narver & Slater, 1990).
Competitor orientation Gaining an understanding of how competitors satisfy their customer’s demands (Narver & Slater, 1990).
Interfunctional coordination Cooperation within and flow of knowledge across the organization (Narver
& Slater, 1990).
Intelligence generation Gaining of an understanding of the needs and preferences of customers and the factors which influence those needs and preferences (Jaworski &
Kohli, 1993).
Intelligence dissemination The spread of market intelligence to relevant departments and employees within the organization (Jaworski & Kohli, 1993).
Responsiveness Responding to generated and disseminated needs and preferences of the market (Jaworski & Kohli, 1993).
Performance Measured as either financial performance (e.g., return on assets or return on equity) or organizational/business performance (e.g., business reputation and customer retention rate).
Moderators: Explanation:
Industry Whether the firms in the sample are operating in manufacturing, service or mixed industries.
Culture Whether the firms in the sample are in a country with a high or low degree of power distance.
Performance Whether performance is measured as organizational/business or financial performance.
Year of data collection Whether the sample data was collected before or after the mean year of data collection 2005.
17 4. Results
4.1 Results of bivariate meta-analysis
Table 3 reports the results of the bivariate meta-analysis of the overall MKTOR and MARKOR measurement scales. The table indicates that the correlation coefficients of both measurement scales are positive and significant, with the linkage being stronger for the MARKOR scale.
Thus, studies which used the MARKOR scale to measure MO received higher effect sizes (p = 0.36, CI = 0.28-0.43) than studies which used the MKTOR scale (p = 0.27, CI = 0.18-0.35).
These findings are in line with several other meta-analyses, including that of Oczkowski and Farrell (1998) and Ellis (2006).
Table 3: Results of MKTOR and MARKOR bivariate analysis
Relationship k N r p SD(p) 95% CI Q test I² #TF Side PTF
Combined 46 9801 .25 .31 .20 .25 .37 424.98*** 89 0 Left .31 MKTOR – P 24 4745 .21 .27 .24 .18 .35 271.62*** 92
MARKOR– P 22 5056 .28 .36 .16 .28 .43 146.04*** 86
Note: k = number of studies included in the analysis, N = total sample size, r = uncorrected correlation coefficient, p = corrected correlation coefficient, SD(p) = standard deviation, 95% CI = confidence interval, P = performance.
Table 4 reports the results of the bivariate meta-analysis for the individual dimensions. The findings indicate that the correlation coefficients of the individual dimensions are all positive and significant at the .001 percent level. Regarding the MKTOR scale, the linkage between the dimension and performance is the strongest for customer orientation (p = 0.31, CI = 0.21-0.40).
With the linkages between competitor orientation (p = 0.26, CI = 0.16-0.36) and interfunctional coordination (p = 0.24, CI = 0.14-0.34) being slightly lower. Regarding the MARKOR scale, the linkage is strongest between responsiveness and performance (p = 0.42, CI = 0.35-0.50), whereas the linkages between intelligence generation (p = 0.33, CI = 0.24-0.42) and intelligence dissemination and performance (p = 0.33, CI = 0.23-0.42) are nearly equal.
The results were tested for publication bias, which is a type of bias that occurs when the studies which were available to be included in the meta-analysis are a biased representation of all existing studies (Hunter & Schmidt, 2004). When there is no publication bias, the #TF column in table 3 and 4 reports ‘0’. When there is publication bias the column reports otherwise. After testing for publication bias, it became clear that none of the results of the measurement scales or individual dimensions suggests publication bias except for the responsiveness dimension.
This may be the case because the mean effect size of responsiveness is relatively large in all
18 studies and there is little variance among the different studies (Hunter and Schmidt, 2004).
Nevertheless, since the difference between the reported p-value and the p-value corrected publication bias reported in column P
TFis relatively small, this will have no real impact on the interpretations of the findings.
To measure the level of heterogeneity the results of the Q statistic and I² are shown. The Q statistic captures the variance between the observed effect and the average effect (Hak et al., 2016). Here, the Q statistic is high and significant for all six dimensions, indicating heterogeneity in the sample. In addition, the I² is used to measure the observed variance which reflects real differences in the effect sizes (Hak, Van Rhee & Suurmond, 2016). It is a measured as a percentage which ranges from 0 to 100. Here, the I² values of the six dimensions range from 88 to 93 and therefore indicate a high level of heterogeneity. Both the Q statistic and I² value thus show that there should be looked for potential moderators.
Table 4: Results of the individual dimensions bivariate analysis
Relationship k N r p SD(p) 95% CI Q test I² #TF Side PTF
MKTOR:
CU - P 24 4745 .24 .31*** .26 .21 .40 324.81*** 93 0 Left .31
CO – P 24 4745 .20 .26*** .28 .16 .36 373.50*** 94 0 Left .26
IC – P 24 4745 .19 .24*** .27 .14 .34 345.34*** 93 0 Left .24
MARKOR:
IG – P 22 5056 .26 .33*** .18 .24 .42 169.08*** 88 0 Left .33
ID – P 22 5056 .26 .33*** .21 .23 .42 230.57*** 91 0 Left .33
R – P 22 5056 .34 .42*** .18 .35 .50 174.01*** 88 2 Left .40
Note: k = number of studies included in the analysis, N = total sample size, r = uncorrected correlation coefficient, p = corrected correlation coefficient, SD(p) = standard deviation, 95% CI = confidence interval, CU = customer orientation, CO = competitor orientation, IC = interfunctional coordination, IG = intelligence generation, ID = intelligence dissemination, R = responsiveness, P = performance.
4.2 Results of the commonality analysis
In this section the results of the commonality analysis will be discussed. A commonality
analysis is a statistical procedure in which the explanation of the variance in the dependent
variable can be decomposed into the unique variance caused by each independent variable, as
well as the variance caused by their common effect. In doing so, the relative importance of the
independent variables comes to light (Seibold & McPhee, 1979). These results will be used to
19 answer the first hypothesis: The effect of MO on firm performance can be explained by the common effect of the dimensions, as well as the unique and bilateral effects.
MKTOR commonality analysis. Table 5 shows the corrected and uncorrected construct correlations of the dependent and independent variables of the MKTOR meta-analysis. Table 6 shows the results of the MKTOR regression analysis. The beta weight column in table 6 indicates the expected difference in the dependent variable by a one increase or decrease of the independent variable when all other independent variables are held constant (Nathans et al., 2012). The r value shows the size and direction of the linear relationship between the dependent and independent variables. In addition, the ‘unique’ column shows the amount of variance caused solely by a specific dimension and the ‘common’ column shows the extent to which the independent variable overlaps with other independent variables in their prediction of the dependent variable (Nathans, Oswald & Nimon, 2012). What becomes clear of this table is that most of the variance in firm performance is caused by the customer orientation dimension (.018), followed by competitor orientation (.006) and interfunctional coordination (.002). This indicates that both competitor orientation and interfunctional coordination need to be paired with customer orientation in order to influence firm performance.
Table 5: MKTOR meta-analytic correlations
Variables 1 2 3 4
1. Customer orientation 1 .47 .52 .24
2. Competitor orientation .47 1 .49 .20
3. Interfunctional coordination .52 .49 1 .19
4. Performance .30 .25 .23 1
Note: Corrected correlation coefficients (p) are presented below the diagonal, uncorrected correlation coefficients (r) are presented above the diagonal. Harmonic mean (N) = 4745.
Table 6: Results of MKTOR regression analysis
b Beta r rs rs2 Unique Common GenDom Pratt RLW
CU .166 .166 .240 .910 .828 .018 .039 .034 .040 .034
CO .094 .094 .200 .758 .575 .006 .034 .020 .019 .020
IFC .058 .058 .190 .712 .519 .002 .034 .016 .011 .015
Total 1.922 .026 .107 .070 .070 .069
Note: r = zero order correlation, rs = structure coefficient, rs² = squared structure coefficient, GenDom = general dominance, Pratt = product measure, RLW = relative weights, CU = customer orientation, CO = competitor orientation, IFC = interfunctional coordination.
20 Table 7 provides a more detailed description of the commonality analysis results. The commonality column displays the commonality coefficient of the dimensions or combination of dimensions and the percentage column describes what percentage of the variance is explained by a specific dimension or combination of dimensions. This table indicates that customer orientation explains 26.5% of the variance in firm performance, competitor orientation explains 8.8%, and interfunctional coordination explains 3.1%. This is where the strength of doing a commonality analysis comes to show. Purely based on the regression analysis displayed in table 6, one could think that because of the low b coefficients of competitor orientation (.094) and interfunctional coordination (.058), these dimensions should not be regarded as relevant influences on firm performance. The results of the commonality analysis in table 6, on the other hand, prove otherwise. The importance of these dimensions becomes visible when they are paired with customer orientation (12.8% and 12.9%) and in the combination of all three dimensions (30.7%). Thus, whilst the unique effect of customer orientation is most important, its effect in combination with one of the other dimensions and in particular the combination of all three dimensions is noticeably stronger. This indicates the importance of the interplay between the individual dimensions of the MO MKTOR scale dimensions.
Table 7: Results of MKTOR commonality analysis
Variables Commonality Percentage
Customer orientation (CU) 0.018 0.265
Competitor orientation (CO) 0.006 0.088
Interfunctional coordination (IFC) 0.002 0.031
CU – CO 0.009 0.128
CU – IFC 0.009 0.129
CO – IFC 0.004 0.052
CU – CO – IFC 0.021 0.307
Total 0.070 1.000
MARKOR commonality analysis. Table 8 shows the corrected and uncorrected construct of the
dependent and independent variables for the MARKOR meta-analysis. Table 9 presents the
results of the MARKOR regression analysis. What becomes clear of these findings is that in
terms of unique effect, the responsiveness dimension is the strongest predictor of firm
performance (.039), followed by intelligence dissemination (.006), and intelligence generation
(.003). Similar as with the MKTOR scale, the unique effect of the individual dimensions is not
21 high enough to make an impact on firm performance and should therefore be seen in combination.
Table 8: MARKOR meta-analytic correlations
Variables 1 2 3 4
1. Intelligence generation 1 .50 .51 .26
2. Intelligence dissemination .50 1 .56 .26
3. Responsiveness .51 .56 1 .34
4. Performance .33 .32 .42 1
Note: Corrected correlation coefficients (p) are presented below the diagonal. Uncorrected correlation coefficients (r) are presented above the diagonal. Harmonic mean (N) = 5056.
Table 9: Results of MKTOR regression analysis
b Beta R rs rs2 Unique Common GenDom Pratt RLW
IG .071 .071 .260 .724 .524 .003 .064 .030 .025 .031
ID .096 .096 .260 .724 .524 .006 .061 .029 .019 .029
R .251 .251 .340 .947 .897 .039 .077 .070 .085 .070
Total 1.945 .048 .202 .129 .129 .130
Note: r = zero order correlation, rs = structure coefficient, rs² = squared structure coefficient, GenDom = general dominance, Pratt = product measure, RLW = relative weights, IG = intelligence generation, ID = intelligence dissemination, R = responsiveness.
Table 10 provides the more detailed description of the commonality analysis results. What becomes clear from the results is that the unique effect of the responsiveness dimension explains 30.1% of the variance in performance caused by market orientation, followed by intelligence dissemination (4.8%) and intelligence generation (2.5%). Overall, these findings are similar to the findings of the MKTOR commonality analysis. Again, purely based on the results of the regression analysis in table 9, one might assume that the intelligence generation and intelligence dissemination dimensions are not an important influence on firm performance. However, looking at their combined importance with responsiveness (12.6% and 15%) proves otherwise.
Again, same as for the MKTOR scale, whilst the unique effect of responsiveness is most
important, its effect in combination with one of the other dimensions and in particular the
combination of all three dimensions is noticeably stronger. Both scales thus confirm the
importance of the individual dimensions as well as the interplay between them. The effect of
MO on firm performance can thus in fact be explained by the common effect of the dimensions,
as well as the unique and bilateral effects. Therefore, the above provides support to H1.
22 Table 10: Results of MARKOR commonality analysis
Variables Commonality Percentage
Intelligence generation (ID) 0.003 0.025
Intelligence dissemination (ID) 0.006 0.048
Responsiveness (R) 0.039 0.301
ID – IG 0.004 0.030
ID – R 0.016 0.126
IG – R 0.019 0.150
ID – I – R 0.041 0.319
Total 0.129 1.000
4.3 Results of the moderator analysis
A moderator analysis was performed in order to answer the second hypothesis: The effects of the dimensions of MO on firm performance will differ for the four different regions; Europe, Asia, the U.S. and Africa. The results of the moderator analysis are reported in table 11.1 and 11.2.
MKTOR region moderator analysis. Table 11.1 reports the results for the MKTOR scale. What
becomes clear of the results in this table is that the linkage between customer orientation and
performance is strongest in Asian (p = 0.48) and European (p = 0.28) countries. These results
are clearly lower for African (p = 0.18) and American (p = 0.18) countries. The linkage between
competitor orientation and performance is strongest in Asian countries as well (p = 0.38),
followed by African (p = 0.27), European (p = 0.16), and American (p = 0.15) countries. For
interfunctional coordination, the linkage is strongest for African (p = 0.40) and Asian (p = 0.37),
followed by European (p = 0.21) and American (p = 0.12) countries. The Q(p)
betweenvalues
show that the differences between the regions are significant for the customer orientation and
interfunctional coordination dimensions (CU Q
between= 11.41, P
between= .010; IC Q
between= 7.93,
P
between= .047), whereas the differences for the competitor orientation dimension are not large
enough to be considered statistically significant (Q
between= 4.09, P
between= .252). In addition,
despite the clear findings for African countries, these findings should not be trusted too much,
as they are based on only two studies.
23 Various reasons can be thought of for the varying results in the different regions. For customer orientation for instance, the consumer markets in and across European and Asian countries may be more heterogeneous than the consumer markets within and across states in the U.S. Thus, the differences between the preferences of Spanish and German customers, for instance, may be bigger than the differences between the preferences of consumers living in South Dakoda and Wyoming. As a result, companies that are active in Asian and European regions will have to deal with a larger variety of wishes among their customers. And in order for these companies to be successful, they will have to put more effort into exploring the needs of their customers and adapting their product or service to these varying wishes in order to be successful. As a result, customer orientation will thus have a stronger effect on firm performance. However, this reasoning is based on the strong assumption that companies mainly sell their products in their own home regions, which is not necessarily the case.
Table 11.1: MKTOR moderator analysis region
Relationship K N p SD(p) 95% CI Q(р)between I² # TF Side PTF
CU – P 24 4745 .31*** .26 .21 .40 324.81*** 93 0 Left .31
Moderator 11.41 (.010)
Europe 6 .28 .29 -.02 .53 92
Asia 7 .48 .16 .35 .59 88
U.S. 5 .18 .18 -.05 .39 85
Africa 2 .18 .10 -.77 .92 31
CO – P 24 4745 .26*** .28 .16 .36 373.50*** 94 0 Left .26
Moderator 4.09 (.252)
Europe 6 .16 .14 -.01 .32
Asia 7 .38 .35 .07 .62
U.S. 5 .15 .19 -.10 .38
Africa 2 .27 .00 -.34 .72
IC – P 24 4745 .24*** .27 .14 .34 345.34*** 93 0 Left .24
Moderator 7.93 (.047)
Europe 6 .21 .14 .02 .39 73
Asia 7 .37 .26 .13 .57 95
U.S. 5 .12 .28 -.20 .42 93
Africa 2 .40 .00 .05 .66 0
Note: k = number of studies included in the analysis, N = total sample size, r = uncorrected correlation coefficient, p = corrected correlation coefficient, SD(p) = standard deviation, 95% CI = confidence interval, CU = customer orientation, CO = competitor orientation, IC = interfunctional coordination, P = performance.
24 MARKOR region moderator analysis. Table 11.2 reports the results of the MARKOR scale. For intelligence generation, the linkage between MO and performance is strongest in European (p
= 0.40) and Asian (p = 0.32) countries, closely followed by American (p = 0.31) and African (p = 0.26) countries. For intelligence dissemination the linkage is strongest in European and Asian countries as well (p = 0.38 and p = 0.34), again followed by American (p = 0.31) and African countries (p = 0.23). For responsiveness on the other hand, the strongest linkage can be found in American countries (p = 0.51), whereas in the other regions the results are highly comparable (p
Europe= 0.41, p
Asia= 0.39, p
africa= 0.37). Nevertheless, the Q(p)
betweenvalues of the various dimensions conclude that the differences across the results of the different regions are not high enough to be regarded as statistically significant (IG Q
between= 4.96, P
between= .175; ID Q
between= 1.69, P
between= .640; R Q
between= 2.00, P
between= .572).
Table 11.2: MARKOR moderator analysis region
Relationship K N p SD(p) 95% CI Q(р)between I² # TF Side PTF
IG – P 22 5056 .33*** .18 .24 .42 169.08*** 88 0 Left .33
Moderator 4.96 (.175)
Europe 6 .40 .11 .29 .51 66
Asia 7 .32 .11 .20 .43 73
U.S. 5 .31 .10 .17 .44 71
Africa 2 .26 .02 -.30 .69 11
ID – P 22 5056 .33*** .21 .23 .42 230.57*** 91 0 Left .33
Moderator 1.69 (.640)
Europe 6 .38 .17 .22 .52 84
Asia 7 .34 .20 .16 .49 90
U.S. 5 .31 .21 .05 .54 92
Africa 2 .23 .13 -.77 .91 80
R – P 22 5056 .42*** .18 .35 .50 174.01*** 88 2 Left .40
Moderator 2.00 (.572)
Europe 6 .41 .17 .26 .54 83
Asia 7 .39 ,11 .29 .47 73
U.S. 5 .51 .23 .25 .71 93
Africa 2 .37 .18 -.87 .97 89
Note: k = number of studies included in the analysis, N = total sample size, r = uncorrected correlation coefficient, p = corrected correlation coefficient, SD(p) = standard deviation, 95% CI = confidence interval, IG = intelligence generation, ID = intelligence dissemination, R = responsiveness, P = performance.
25 Based on these findings can thus be confirmed that the effects of the dimensions of MO on firm performance differ for the four different regions; Europe, Asia, the U.S., and Africa, with the findings being statistically significant only for the customer orientation and interfunctional coordination dimensions. Therefore, the above provides partial support to H2.
Robustness check. In this region moderator analysis two Russian and one Turkish study were included. Both countries belong to Europe and Asia and were categorized based on the geographic location of their capital. Therefore, the Russian studies were included as belonging to the European region and the Turkish study as belonging to the Asian region. To ensure that this decision did not have major influence on the results, a robustness check was conducted. In this check, the countries were categorized the other way around to see how this influences the findings. The conclusion from this check is that categorizing the countries the other way around only leads to a minimal change in the results (e.g., CU Q
between= 11.41, P
between= .010 changed to Q
between= 11.69, P
between= .009 and ID Q
between= 1.69, P
between= .640 changed to Q
between= 1.96, P
between= .580). Therefore, the original categorization of the two countries was remained.
To further explain the variance in the effect of the dimensions on firm performance, several other moderators will be tested as well. The moderators included in the analysis are industry type (i.e., service, manufacturing, or mixed; table 12.1 and 12.2), culture (i.e., high or low power distance countries; table 13.1 and 13.2), performance measure (i.e., organizational or financial performance; table 14.1 and 14.2) and year of data collection (i.e., before or after 2005; table 15.1 and 15.2).
MKTOR industry moderator analysis results. In general, table 12.1 shows that engaging in MO
is positive for firm performance in all industries. For customer orientation, the findings show
that the linkage between customer orientation and performance is strongest in manufacturing
industry’s (p = 0.34) as compared to service (p = 0.28) and mixed industries (p = 0.30). For
competitor orientation the linkage with performance is strongest in service industries (p = 0.28),
as compared to manufacturing (p = 0.26) and mixed industries (p = 0.24). Finally, for
interfunctional coordination the linkage with performance is the strongest in manufacturing and
mixed industries (p = 0.26), as compared to service industries (p = 0.20). Overall, all three
coefficients are highest for customer orientation. Despite the slight difference between the
results for the different industries, for all three dimensions these differences are not big enough
to be regarded as statistically significant (CU Q
between= 0.37, P
between= .832; CO Q
beween= 0.04,
P
between= .982; IC Q
between= 0.38, P
beween= .826)
26 Table 12.1: MKTOR moderator analysis industry
Relationship k N p SD(p) 95% CI Q(р)between I² #TF Side PTF
CU - P 24 4745 .31*** .26 .21 .40 324.81*** 93 0 Left .31
Moderator 0.37 (.832)
Service 7 .28 .28 .07 .46 94
Manufacturing 8 .34 .18 .19 .48 87
Mixed 9 .30 .34 .06 .51 95
CO – P 24 4745 .26*** .28 .16 .36 373.50*** 94 0 Left .26
Moderator 0.04 (.982)
Service 7 .28 .37 .01 .51 97
Manufacturing 8 .26 .21 .08 .42 90
Mixed 9 .24 .29 .04 .43 93
IC – P 24 4745 .24*** .27 .14 .34 345.34*** 93 0 Left .24
Moderator 0.38 (.826)
Service 7 .20 .19 .04 .34 88
Manufacturing 8 .26 .34 .00 .49 96
Mixed 9 .26 .29 .06 .45 93
Note: k = number of studies included in the analysis, N = total sample size, r = uncorrected correlation coefficient, p = corrected correlation coefficient, SD(p) = standard deviation, 95% CI = confidence interval, CU = customer orientation, CO = competitor orientation, IC = interfunctional coordination, P = performance.
MARKOR industry analysis results. The results in table 12.2 indicate that the correlation
coefficients of the three industry types are highest for the responsiveness-performance
relationship (p
service= 0.45, p
manufacturing= 0.49, and p
mixed= 0.34). The linkage between
intelligence generation and performance is the strongest in service industry’s (p = 0.43),
followed by manufacturing (p = 0.38) and mixed industry’s (p = 0.18). For intelligence
dissemination, the linkage with performance is strongest in service industry’s as well (p = 0.43),
followed by manufacturing (p = 0.35) and mixed industry’s (p = 0.22). The Q(p)
betweenvalues
show that the differences between the industries are only significant for the relation between
intelligence generation and firm performance (Q
between= 8.64, P
between= .013). For the other two
industries the results are slightly insignificant (ID Q
between= 4.55, P
between= .103; R Q
between=
3.66, P
between= .160).
27 Table 12.2: MARKOR moderator analysis industry
Relationship k N p SD(p) 95% CI Q(p)between I² #TF Side PTF
IG – P 22 5056 .33*** .18 .24 .42 169.08*** 88 0 Left .33
Moderator 8.64 (.013)
Service 7 .43 .17 .25 .59 86
Manufacturing 8 .38 .10 .28 .46 62
Mixed 7 .18 .19 -.02 .36 91
ID – P 22 5056 .33*** .21 .23 .42 230.57*** 91 0 Left .33
Moderator 4.55 (.103)
Service 7 .43 .21 .23 .59 90
Manufacturing 8 .35 .28 .15 .52 92
Mixed 7 .22 .17 .04 .39 90
R - P 22 5056 .42*** .18 .35 .50 174.01*** 88 2 Left .40
Moderator 3.66 (.160)
Service 7 .45 .22 .24 .61 91
Manufacturing 8 .49 .18 .36 .60 83
Mixed 7 .34 .16 .19 .47 88
Note: k = number of studies included in the analysis, N = total sample size, r = uncorrected correlation coefficient, p = corrected correlation coefficient, SD(p) = standard deviation, 95% CI = confidence interval, IG = intelligence generation, ID = intelligence dissemination, R = responsiveness, P = performance.