Meta-analysis Study of Marketing Innovation on SME Business Performance in Ethiopia

This study looked at how market innovation affected SMEs' performance in Addis Abeba, Ethiopia, with the use of government support programmes as a moderator. The study's meta-analysis will be guided by the Schumpeter innovation theory, which also incorporates the diffusion of innovation theory, expectancy theory, institutional theory, stakeholder theory, absorptive capacity theory, resource-based view/theory, dynamic capability theory, R-A theory, unified theory of acceptance and use of technology. The researcher used an effect size approach based on a forest and funnel plot to scan, exclude, and include relevant material. Researchers discovered a connection between the Dimension of Innovation and company performance based on an extensive literature review. Also Based on the studied literature, the researcher discovered a link between the moderation of the government support program and the performance of SME firms in the area of innovation. An exploratory sequential mixed research design will be used to carry out this study. The study attempts to create a conceptual framework and testable hypotheses based on the current literature. It was discovered that the impact of marketing innovation on a firm's performance was moderated by government support programs. Businesses must be encouraged to adopt government support programs with a moderating influence as a result.

According to the study's findings, the leadership of the owners, access to infrastructure, entrepreneurial training, entrepreneurial mentality, and government assistance all had a major impact on how innovative service and manufacturing MSEs were (Bansal & Kant. 2018.). The performance of Mozambique SMEs' exports is positively impacted by their innovation skills (Moreira & Navaia, 2022). Using sustainable strategies including green HRM, a green supply chain, green innovation, and green marketing, Alraja et al. (2022) discovered a substantial correlation between technical innovation and sustainable performance. According to Adam & Alarifi (2021b), SMEs' innovation methods have a substantial impact on their performance and ability to survive, and it is crucial to have external support to further enhance this impact. Government Support significantly improved SME performance and attenuated the association between IC and SME performance in a favorable way (Otache & Usang, 2021).
The performance of manufacturing SMEs is positively and significantly impacted by marketing innovation. Both organizational innovation and innovation culture have a significant and advantageous effect on business performance. The study's findings indicated that product innovation had a favorable impact on consumer interest (Sinaga et al., 2021). According to this study, there is an association between the performance of SMEs and SI (strategic innovation). They discovered a robust, positive association between government funding for innovative techniques and the performance of SMEs (Adam & Alarifi,2021). The performance of SMEs is strongly correlated with each of the characteristics (Bansal & Kant. 2018.). Government assistance plays a larger role as a moderator between the acceptance of technical innovation in the context of the environment, where there is a positive relationship and the rate of change.
On the other hand, research has shown that both market-driven and market-driving innovations have a major impact on a firm's performance. The level of competition and the state of technology considerably mitigate their effects. Variables relating to innovations and financial performance have a negative correlation (Peng et al., 2021). Using time-series data, according to (Edeh et al., 2020). First, we discover that while process innovation boosts export performance, product innovation has a detrimental effect. On the other hand, marketing innovation had a negligible and minor impact. A firm's performance, both financially and non-financially, is not considerably impacted by aggregate innovation. Only the marketing innovation of the four dimensions of innovation has a substantial impact on the financial and non-financial success of the company (Mabenge et al., 2023). The true impact and significance of marketing innovation for manufacturing organizations are not presented (Del Carpio Gallegos & Miralles, 2020).

Specific Objectives
➢ Marketing innovation has a statistically significant relationship with firm performance. ➢ Government support program has a moderate effect between marketing innovation and firm performance.  Figure 1. Forest plot Researchers by the above forest plot discovered through met analysis that the top of the plot's x-axis represents the effect size scale of the examined systematic literature. Except for the bottom row, each row shows the estimated effect size from a reviewed systematic study as a point and (95%) confidence interval. The results of a single study were presented in this statistically accurate manner as an estimate of the interval in which the "actual" effect (of the studied systematic literature) was most likely to lie. Every study included in the meta-analysis was thought to be a study of a complete probability sample of a particular population, according to researchers. A smaller or larger bullet in the forest plot corresponds to the point estimate. A smaller or larger bullet in the forest plot corresponds to the point estimate. The proportional size of these bullets indicates how important a study was in producing the meta-analytic outcome.

Empirical Literature Review
Source: Meta Essentials (2023)   (2023) Researchers discovered that the moderator of the government support program is a third variable that influences the relationships between the other two. Since the relationships between two variables are represented by their effect sizes, any variable that predicts the effect sizes is a moderator. The significance of the interaction term was the main consideration for the researcher when evaluating the findings of a moderation analysis. The moderator Government support program has a considerable moderating effect on the relationship between market innovation and business performance, according to research that found the interaction term's effect on the endogenous construct to be significant.
T2 was significant, and the researchers used this information to estimate the variance of the real impact sizes. While computing the variance of these effects, researchers assumed that "if we had an indefinitely large sample of studies, each itself infinitely big (such that the estimate in each study equaled the genuine effect), this variance would be τ2."  In our meta-analysis, the between-study variation is 2. It is an estimation of the genuine effect sizes' underlying distribution variance. As the chart above demonstrates, there are several suggested methods to calculate τ2.

Publication Bias Analysis
According to researchers, an area of study's body of research is likely to be prejudiced in many different ways. The likelihood that a statistically significant result will be published is predicted to be higher than that of a statistically non-significant result. As a result, the study's estimated cumulative effect size may be higher than it is. The examination of publication bias aims to (1) alert the reader to this potential publication bias and (2) correct the estimate for the total effect magnitude.
Source: Meta Essentials (2023) Figure 3. Funnel plot Six distinct analyses used by the researchers to point to publication bias are provided by Meta-Essentials. A funnel plot is a type of analysis. It is believed that observed effect sizes should be more or less symmetrically distributed around the total effect size when measured with similar precision (i.e., with similar standard error). As was already indicated, it is predicted that outcomes farther from the null will outnumber those closer to it. The figure above shows that this is not the case. According to the Trim-and-Fill approach, there are no imputed data points, hence the funnel plot shows there is no asymmetry in the distribution of effect sizes. However, the Trim-and-Fill strategy would impute one or more studies and afterward modify the overall effect size for the potentially missing studies if we found asymmetry.   (2023) The researchers employed Egger's regression test to quantitatively evaluate this disparity. It looks at the correlation between the measured effect sizes and their sample standard errors (SEs); a large correlation shows the presence of effects from small studies. With a p-value of 0.775, Egger's test for a regression intercept revealed no indication of publication bias. The funnel plot suggests that there may be publishing bias. The rank correlation test by Begg and Mazumdar produced a p-value of 0.091, suggesting potential publication bias.  (2023) The researchers observed that considerable heterogeneity ranged from 50% to 90%. When there is substantial statistical heterogeneity, it means that the studies are not all estimating the same quantity. However, this does not necessarily imply that the true intervention effect varies. Significant statistical heterogeneity arose from methodological diversity or differences in outcome assessments.
Source: Meta Essentials (2023) The standardized residual histogram is based on the researchers' hypothesis that a normal distribution should be expected to surround the combined effect size for the z-scores of different studies, also known as standardized residuals. Researchers binned the residuals and plotted them against an expected normal distribution to see whether there are any outliers in the effect sizes. The proportion of residuals in each of the nine bins, which are used to organize the standardized residuals, determines the height of the bar ( Figure 5).
Source: Meta Essentials (2023) Figure 5. Galbraith plot To produce the Galbraith plot or radial plot, researchers must first conduct an unweighted regression of z-scores on the inverse of the standard error with the intercept restricted to zero (Galbraith, 1988). (see Figure 24). To find outliers in the effect sizes, use this figure. The two (lighter-colored) confidence interval lines are expected to contain 95% of the study's results. A map, a table with regression estimates, and a table containing studies are all provided by MetaEssentials.  (2023) Researchers have also employed normal quantile plots, often known as Q-Q plots, to determine whether data are normally distributed. The researchers anticipated that all data points would be roughly on a straight line, indicating that the data would be dispersed according to a conventional normal distribution. A table comprising studies, a graphic, regression estimates, and an input option for calculating sample quantiles make up this component of Meta-Essentials. The calculated normal quantile, sample quantile, and research names are shown in the table. These normal and sample quantiles are shown on the plot along with a regression line. The user has the choice to base the sample quantiles on "Standardized residuals" or "Z-scores" using the input option.

Failsafe-N Tests
Many estimates of the Failsafe figures are included in the Publication Bias Analysis sheet's last section. To demonstrate this, researchers will pretend that several other papers are never published for any given study. Suppose that the results of these extra studies are negligible, or that their impact sizes are close to zero. The failsafe number then calculates the approximate number of such extra studies needed to make the combined effect size from the included and additional studies inconsequential, or nearly zero.

Conclusion
To assess the (weighted) average effect size, the dispersion of effect sizes, the homogeneity (or heterogeneity) of the entire set of observed effect sizes and of subgroups, and to explore the applicability of possible moderators, researchers performed a meta-analysis. The degree of heterogeneity should be evaluated and analyzed before any judgments are made. Only when there is no question about the homogeneity of a group or subgroup of observed effect sizes Normal quantile plot may "combined" effect sizes be used as an outcome, and even then, only for the domain that is defined by this particular group of populations. The main outcome of most meta-analyses is an understanding of the dispersion of genuine effects because relevant heterogeneity is typically discovered by the researchers in this study. Meta-analysis serves as a tool for developing theories concerning "moderators" of the effect under those circumstances. The meta-analysis shouldn't be utilized for "testing" or for making generalizations about the magnitude of an effect throughout the entire domain or in areas of the domain that haven't been thoroughly studied.

Conflict of Interest
No conflict of interest