Systematic review of meta-analysis studies in the tourism and hospitality literature

Referred to as "the analysis of the analyses", meta-analysis has become a popular research area in the tourism and hospitality literature, as well as in many different fields of science. This great interest in meta-analysis has led to many meta-analysis studies in the tourism and hospitality literature; however, no general systematic perspective has been established for these studies. This paper examines the meta-analysis studies published in the tourism and hospitality literature via various parameters and presents a comprehensive classification. Because this paper uses a theoretical scope to systematically scrutinize meta-analysis studies conducted in the tourism and hospitality literature, it not only presents state-of-the-art research, it can also become a roadmap and reference for future researches.


Introduction
Science refers to the total of cumulative efforts that comprise a combination of research methods conducted in certain disciplines. Systematically reviewing existing research helps ensure this integration and eliminates the necessity of "constantly reinventing the wheel" (Gretzel & Kennedy -Eden, 2012, p.459).
The empirical results achieved in individual studies on the same research subject can often be inconsistent with each other and are insufficient to provide general conclusions about the subject being studied (Hunter & Schmidt, 2014, p.18). Thus, meta-analysis is based on the idea of integrating evidence from a range of different sources in order to provide reliable answers to important questions. Metaanalysis can be defined as the process of systematically gathering the findings of research, synthesizing them according to certain principles, and analysing them collectively (Shelby & Vaske, 2008, p.98). Meta-analysis represents the combination of scientific efforts that are carried out independently of each other (Parker et al., 2013, p.713).
The tourism and hospitality field is a relatively young area that involves several disciplines, methodologies and subjects. The complex nature of the structure of information in the field of tourism and hospitality makes it difficult to generalize research findings. Meta-analysis brings different results obtained in tourism and hospitality research together under one roof and gives the results a reliable and generalizable structure (Gretzel & Kennedy -Eden, 2012, p.466). Meta-analysis objectively analyses individual studies that are found with a comprehensive literature search within the framework of structured and standardized techniques. The findings obtained from meta-analysis studies help reveal the biases of individual studies, as well as their strengths and weaknesses (Russo, 2007, p.637).
In the existing literature, no research was found on the general development, history, and stereotypes of meta-analysis studies conducted in the field of tourism and hospitality. As the first scientific initiative to fill this gap in the literature, this paper aims to identify the meta-analysis studies carried out in the field of tourism and hospitality with a comprehensive literature review and to analyse these studies in a systematic framework. Examining these studies from a systematic perspective will provide an overview of the historical development, research themes, trends, effects, and intellectual structure of metaanalysis in the field. In addition, this paper, which filters a vast literature for researchers interested in the subject, is a guide to determining research gaps and opportunities in this field by examining metaanalysis studies in the light of various parameters. This paper explains the generally-accepted methods of research synthesis and proposes the importance of meta-analysis in the context of research synthesis. In the following sections, general information on meta-analysis is presented and classified in terms of research design. Then, the paper presents the scope of the meta-analysis studies conducted in the tourism and hospitality literature and an examination of these meta-analysis through various parameters. Finally, the paper discusses findings in terms of their effects and contributions to the tourism and hospitality literature.

Literature Review Research Synthesis Methods
Research synthesis studies can be defined as studies that summarize the research conducted on a specific discipline or subject and that provide important contributions to the development of knowledge (Kim, Bai, Kim & Chon 2018, p.49). Although there are several different typological approaches to research syntheses (Grant & Booth, 2009), there are three general approaches: narrative review, systematic review and meta-analysis.
Narrative review refers to studies that examine two or more studies published on a certain subject and synthesize their findings, results and evaluations (Neely et al., 2010, p.6). In a narrative review, the researchers review existing research and summarize the findings. The findings of a narrative review are based on comparing studies with statistical significance and those with insignificant findings. Manuscripts prepared by people who are experts in their field are accepted as "expert opinion" and evaluated as the lowest level of evidence because of concerns regarding potential subjectivity in the selection and interpretation of studies (Quintana & Miami, 2006, p.839-840).
Systematic review is the process of defining and evaluating many studies on a subject through a used systematic methodology. A systematic review fits empirical evidence to conformity criteria determined beforehand in order to answer a certain research question (Haidich, 2010, p.30). This type of review presents a transparent and impartial examination of existing information (Neely et al., 2010, p.6). When statistical methods are applied to a systematic review, it is referred to as a meta-analysis (Koretz & Lipman, 2017, p.317). Quantitative meta-analysis statistically summarizes and interprets the theoretical results obtained from a systematic review (Quintana & Miami, 2006, p.839).
Since it was initially defined by Glass (1976), meta-analysis has gradually become a popular tool in evidence-based decision-making. Meta-analyses studies generally occupy the topmost place in the evidence hierarchy (Parker et al., 2013, p.713). Meta-analysis produces objective evidence by reducing the subjective bias through the standardization it provides in the process from collecting individual studies in the sample unit to the statistical presentation of the analyses (Zhang et al., 2014, p.214). By combining the results of individual studies with statistical methods, more accurate and reliable estimates of independently reported effect coefficients can be achieved. Meta-analysis has far greater power and scientific rigor in revealing real correlations because of the accumulation of estimated effect sizes (Garg, Hackam & Tonelli, 2008, p.253-254).

Meta-analysis
Since Gene Glass first used the term "meta-analysis" at the annual congress of the American Educational Research Association in 1976, meta-analysis has become popular in many different disciplines. The fact that meta-analysis reveals statistically stronger, reliable, and accurate findings increases its popularity (Shelby & Vaske, 2008, p.96). Meta-analysis is a method of research involving the analysis of analyses, as its name indicates (Gretzel & Kennedyn -Eden, 2012, p.459). Meta-analysis is based on the fundamental values of scientific endeavour, as it involves quantitative methods, relationality and causality (Borenstein, Hedges, Higgins & Rothstein, 2009). The power of meta-analysis stems from its provision of reliable results through combining the results of individual studies that may be individually inadequate in terms of statistical significance (Russo, 2007, p.637).
As in any kind of research, the success of a meta-analysis study depends on the quality of data, careful coding and analysis and the complete understanding of the statistical assumptions (Gretzel & Kennedy-Eden, 2012, p.464). Along with being more valuable than a single study (Haidich, 2010, p.36), metaanalytical approaches are as good as the individual studies on which they are based and they enable researchers to consider the whole picture (Shelby & Vaske, 2008, p.105-106). Meta-analysis identifies differences in research findings and provides empirical support for how these differences originate. It also offers the possibility to define various factors that may affect the research findings as a moderator or subgroup (Quintana & Miami, 2006, p.839).
Meta-analysis studies focus on effect sizes and confidence intervals. The parameter of effect size addresses the actual relationship between two variables in a population. The confidence interval, in contrast, is the expected interval of the effect size at a specific level of confidence (Quintana & Miami, 2006, p.840-841).
Because meta-analyses integrate a collection of different studies, an element of heterogeneity is often observed in meta-analyses (Parker et al., 2013, p.713). The most prevalent statistical test used to measure heterogeneity is the Q test. According to whether the studies incorporated into the meta-analysis exhibit homogenous or heterogeneous structures, the mathematical combination of the results can be actualized under the assumption models of "fixed" effects or "random" effects (Garg et al., 2008, p.257). If the individual studies incorporated into the meta-analysis exhibit a homogenous structure, the fixedeffects model should be chosen for the meta-analysis, whereas if they have heterogeneous structures, the meta-analysis should use the random-effects model (Neely, et al., 2010, p.7;Celiker, Ustunel & Guzeller, 2019, p.5). The fixed effects model argues that all individual studies included in the metaanalysis have a common (true) effect size. Therefore, all factors that are likely to differentiate the actual effect size are the same in all the individual studies included in the analysis. This means that the real effect size is the same in all the studies (Borenstein et al. 2009, p.63) The random effects model emphasizes that the actual effect sizes of individual studies on different universal communities will differ (Neely et al., 2010, p.10).
Despite the important advantages of meta-analysis discussed above, there are certain limitations of this method; meta-analysis is not a key solution for everything. Meta-analysis, especially in the social sciences, is a relatively new methodology that is still early in development and is contentious because of potential biases and the risk of comparing things that are incomparable (Gretzel &Kennedy-Eden, 2012, p.459).
In meta-analysis studies, factors such as the violation of meta-analytical methods, errors in the selection process of the studies to be incorporated into the analysis, publication bias and small sample sizes can cause erroneous inferences regarding the relationships between variables. One of the most important negative aspects of meta-analysis is the possibility that the studies included in the analysis may contain errors or bias. It may not be appropriate to include different studies in meta-analysis in terms of measurement models, measurement types, measurement tools, and sampling characteristics. Another criticism is that a meta-analysis only incorporates published studies into the sample and that unpublished studies are ignored (Shelby & Vaske, 2008, p.106). Publication bias can occur if the researches included in a meta-analysis on a particular subject do not represent the population of studies conducted but not published on the same subject (Rothstein, Sutton & Borenstein, 2005, p.1). The biases of the studies incorporated into a meta-analysis can be tested using publication bias techniques such as a Funnel plot, Classic fail-safe N, Begg and Mazumdar Rank Correlation, Egger regression and Duval Tweedie's trim-and-fill (Celiker, Ustunel & Guzeller, 2019, p.11).

Meta-analysis According to Research Designs
Meta-analysis studies can be classified in three different ways in terms of research designs. These include correlation studies examining the relationships between the effect sizes of the variables, experimental studies on the differences between the experimental and control groups, and regression (meta-regression, etc.) studies evaluating the relationship between the variables of study level and effect size (Hedges & Olkin, 1985;Borenstein et al., 2009;Hunter & Schmidt, 2014).

Correlational Meta-analysis
Correlational meta-analysis aims to determine the true strength and direction of the relationship between the two measurements (Pigott, 2012, p.11). In this context, they examine the correlation coefficients (effect sizes) reported in individual studies on the relationships between two continuous variables (Neely et al., 2010, p.10-11). In meta-analysis studies based on correlation, the correlation coefficients (r) obtained from individual studies are first converted into Fisher's Z scale (Borenstein et al., 2009, p.41). According to these transformed values, average correlation and confidence interval calculations are performed, and the results are converted back into correlation and reported (Pigott, 2012, p.11).

Experimental Meta-Analysis
Experimental meta-analysis compares the mean scores of two independent or matched groups (experimental and control) (Borenstein et al., 2009, p.20). In experimental studies, the fundamental statistic is not the correlation coefficient but the standardized difference between two groups (experimental and control groups). This is the difference between the two means in standard deviation units and it is referred to as the d-value statistic (Hunter & Schmidt, 2014, p.62). The standardized mean difference (d) is the effect size index, which reveals differences between two independent groups, such as the experimental and control groups (Pigott, 2012, p.10). In addition, if the summarized data is based on a binary result for two groups, the suitable effect size will usually be a risk ratio, odds ratio or risk difference (Borenstein et al., 2009, p.18).

Regression-based Meta-analysis
As we use regression or multiple regression to examine the relationship between the independent variable(s) and a dependent variable, we can perform regression-based meta-analysis by defining individual studies as variables and determining the effect size as the dependent variable. In regressionbased meta-analysis, the meta-regression model is commonly used. Meta-regression involves designating a weight to each study and choosing the suitable model (fixed-effect or random-effects). Further, the R 2 index, which is used to measure the variance ratio revealed by the covariables, needs to be modified in order to be used in meta-analysis (Borenstein et al., 2009, p.187-203). Meta-regression is the suggested strategy when there is a sufficiently large effect size, because it has the power to examine the effect of more than one variable on the effect size (Pigott, 2012, p.25).

Meta-analysis in Tourism and Hospitality Literature
The tourism and hospitality field has a considerable research archive in theoretical and methodological terms and the synthesis and critical evaluation of approaches has become a necessity. The approaches in the existing literature within this field have been carried out in the scope of different research themes (Gretzel & Kennedy-Eden, 2012, p.469).
Meta-analysis studies in the tourism and hospitality literature often focus on specific areas of interest, such as the tourism economy, recreation, eco-tourism, perceptions, attitudes, behaviour and behavioural intentions. Especially common are meta-analysis studies examining demand estimation, demand flexibility and tourist incomes (Crouch, 1994(Crouch, -1995(Crouch, -1996Baaijens, Nijkamp & Montfort, 1998;Lim, 1999;Peng, Song & Crouch, 2014). In one of these, Crouch (1994) combined the experimental findings of 80 studies on international tourism demand in order to ascertain whether there is a systematic difference between long-term and short-term tourism. In the study, the income elasticity of demand, price elasticity, currency elasticity, transportation cost-awareness and the promotion expenditure-awareness of demand were examined and the growth rates of long-term travel were compared with those of short-term travel. Baaijens et al. (1998) focused on regional tourist multipliers, which measure the effect of tourism expenditure on the economy of a specific region and the differences between tourist income multipliers were examined on the basis of various studies carried out on tourist multipliers. Lim (1999) examined the 70 studies aiming to model international touristic demand within the scope of the relationships between international tourism demand, income, transportation costs and tourism prices. Kim and Schwartz (2013) examined the relationship between tourism demand and data characteristics using a meta-analytical approach. The findings suggest that data characteristics can provide reliable indicators for which models are more accurate for which tourism estimation methods. In addition to economy-focused meta-analysis studies, there are meta-analysis studies on recreational experiences, opportunities, activities and leisure time motivations (Manfredo, Driver & Tarrant, 1996;Cisneros-Montemayor & Sumaila, 2010;Sato, Wood & Lindenmayer, 2013;Wang, 2013) as well as on the effects of tourism and tourist activities on various species in nature and wildlife (Anderson, Rocliffe, Haddaway & Dunn, 2015;Bateman & Fleming, 2017;Vaz et al., 2018). For instance, Pierskalla et al. (2004) combined individual studies on recreation opportunities using meta-analytical techniques. Sato et al. (2013) examined the effects of winter recreation on faunal life and concluded that in areas affected by winter recreation, fauna variety is much lower than in natural areas.
In the tourism and hospitality literature, there are meta-analysis studies addressing the relationship between destination image, customer satisfaction and customer loyalty (Zhang, Fu, Cai & Lu, 2014;Dolnicar, Coltman & Sharma, 2015;Tanford & Jung, 2017), as well as customer satisfaction and the antecedents and consequences of customer loyalty (Ladeira, Santini, Araujo & Sampaio, 2016;Tanford, 2016). Zhang et al. (2014), examined the relationship between destination image and tourist loyalty and revealed that the general image has the most significant effect on tourist loyalty, followed by the affective image and cognitive image. Nghiêm-Phú (2018) examined the relationship of perceived destination qualities with satisfaction and determined that a destination's image, quality and attributes have a positive effect on tourists' overall satisfaction. Ladeire et al. (2016) aimed to specify the dimensions of satisfaction by evaluating its main antecedents and consequences. The findings obtained in the study show that satisfaction is influenced by factors such as quality, destination image, environment, perceived hedonic benefit, and monetary value. However, it has been concluded that satisfaction has important effects on individuals' purchasing intentions, loyalty, and trust levels. Tanford (2016) presented a theoretical framework for classifying the antecedents and consequences of hospitality loyalty and measured these components using a meta-analysis. The researcher presented the strong links between direct antecedents of loyalty (satisfaction, emotional commitment, service quality and trust) and general loyalty. In another meta-analysis, Tanford and Jung (2017) evaluated the factors contributing to festival satisfaction and loyalty. In the study, it was concluded that the festival (which includes the program, entertainment, and thematic activities) and the environment (including atmosphere, comfort, and facilities) are the most important factors in ensuring satisfaction and loyalty. As mentioned above, studies carried out on perception and behaviour have generally focused on tourists. Meta-analysis studies addressing the perceptions and behaviour of tourism and hospitality industry workers are limited. In a study conducted within this scope, Celiker et al. (2019) discussed the relationship between the emotional labour and burnout levels of tourism and hospitality industry workers.

Methodology
A systematic analysis is a clearly formulated inquiry that uses systematic and clear methods in order to define, select and critically evaluate the research within specific scopes, as well as collect and analyse the data obtained from the studies involved in the analysis process (Crocetti, 2016, p.3). The process of a systematic review should involve a comprehensive literature review, a declaration of the criteria of inclusion and exclusion for the defined studies, the application of a standard protocol to evaluate the studies, a summary of findings and suggestions for future studies (Koretz & Lipman, 2017, p.316).

Data Collection
The data collection procedure of this current paper comprised two stages. First, a search was carried out with the keywords "meta-analysis", "meta-analytic" and "meta-review" in two databases: Web of Science and EBSCOhost Hospitality & Tourism Complete. There are several justifications for selecting these two databases as the primary data collection tools. EBSCOhost Hospitality & Tourism Complete is one of the most important databases indexing research focused on tourism and hospitality. Web of Science is a comprehensive scientific search engine containing many databases (Emerald, JSTOR, Sage Journal, Springer-Link, Taylor & Francis, etc.) that indexes the journals focused on tourism and hospitality in a separate academic category. Taking data from the Journal Citation Report 2018 as a basis and excluding journals in the area of sports sciences out of the 50 journals indexed in the research category of Hospitality, Leisure, Sport & Tourism in the Web of Science, 51 studies were found from the search of 25 tourism and hospitality-focused journals. The search of EBSCOhost Hospitality & Tourism Complete returned 105 studies after excluding duplicates of the publications accessed in the previous search.
Second, a search was carried out using the Scopus database and Google Scholar search engine. These two search sources were selected in order to have a more comprehensive search result. Scopus is the most extensive refereed literature, abstract and citation database (Elsevier, 2019). Google Scholar is considered the most comprehensive search engine, with 389 million records (Gusenbauer, 2019, p.177). Furthermore, because Web of Science only contains those journals that are involved within the scope of SSCI and ESCI (Emerging Science Citation Index), a general search was needed to access studies published outside the scope of these journals. In order to exclude the meta-analysis studies applied to different fields and to prevent wasting time on unnecessary research, the "tourism" and "hospitality" keywords were added to the existing keywords. As a result of the review, after excluding the studies obtained from the databases in the first stage, 36 studies were accessed in the Scopus database and 21 studies were accessed using Google Scholar, thereby ensuring a more reliable and comprehensive literature review. In total, 213 studies were obtained for the literature review.

Inclusion Criteria
The 213 studies obtained for the literature review were evaluated for compliance with predetermined inclusion criteria. Accordingly, published meta-analyses that were the correct type, in English, conducted in the discipline of tourism and hospitality and had available full-text were included in the analysis unit. This paper had no year-interval limitation. As of the last date in which the literature review was carried out (June 2019), all studies meeting the inclusion criteria were incorporated into the sample. In the paper, the inclusion flowchart adapted from Moher, Liberati, Tetzlaff and Altman (2009) was applied and the process was completed by excluding ineligible studies for each criterion. Accordingly, studies that were outside the article type (24), were not in English (11), had inaccessible full-text (12), did not address the tourism and hospitality field (7) and did not involve meta-analytical methods in terms of literature review/narrative review/systematic review (113) were excluded from the analysis. Finally, a total of 46 meta-analysis studies were included in the analysis (Table 1).

Coding
The studies to be included in the systematic analysis were examined according to the coding criteria determined by the researchers. In the coding form, the studies were primarily coded on the basis of their definitive properties (titles, years of publication, author(s), the countries in which they were conducted and journals in which they were published) and data collection channels and number of samples. The authors coded on the basis of the countries of the affiliated institutions in which the researchers carried out their research activities, as opposed to their countries of citizenship. When two different authors of one study were from the same country, the country was counted only once. Furthermore, since changes in the institutions may also result in country-related changes, calculations were finalized taking this matter into account. Because authors can conduct research simultaneously in different institutions, when they indicated two different countries in a single study, only the first country indicated was counted for the coding.
Subsequently, the studies were classified according to the research areas and designs and examined with regard to the applied meta-analysis model and publication bias tests. During the last stage, information regarding the citation numbers of the studies was coded.
The research area themes from Park, Phillips, Canter and Abbott (2011) and Kim et al. (2018) were used for categorizing the studies' research areas, though reorganized in a way to fit the studies included in our research's dataset. Within this scope, the studies were categorized in seven different topic titles:

Development of theories, models and scales (D_TMS), Environmental and cultural issues (E&C), Economic and financial aspects (E&F), Methodologies and statistics techniques (M&S), Marketing strategies and practices (MS&P), Specific sectors of tourism, (health tourism, event tourism, outdoor recreation, leisure, marine, cruises (TOUR_SEC) and Tourist and visitor attitude, value, satisfaction, behaviour and behavioural intention (T&V Beh).
Studies that focused on more than one research area matter were classified on the basis of the main research area theme.
The studies were evaluated in three categories according to their research design: correlational, experimental and regression studies. According to the meta-analysis model applied, the studies were examined on the basis of two different models: fixed-effect model for when the sample was homogenous or random-effects model for studies with a heterogeneous structure.
Citation analysis data from Google Scholar and Web of Science were used for coding the citation analyses of the studies. For ranking the general citation numbers of the studies, Google Scholar, which includes more comprehensive citation data, was taken as a basis. The citation data from the Web of Science, in contrast, gives the number of citations of the studies within the scope of SSCI. The coding was carried out by three researchers separately and a consensus was reached on both the inclusion of the studies and all coding.

Publication Numbers According to Years and Periods
The first meta-analysis study within the discipline of tourism and hospitality was carried out by Geoffrey I. Crouch in 1994.  (Table 2). Examining the studies for co-authorship showed that study with the highest number of authors was "An indicator-based approach to analyse the effects of non-native tree species on multiple cultural ecosystem services", published in 2018 with 11 co-authors. Among the 46 studies, 13% were single-author (n=6), 30.4% were two-author (n=14), 28.3% were three-author (n=13), 21.7% were four-author (n=10), 4.3% were five-author (n=2) and 2.2% were eleven-author studies (n=1).
The studies were carried out by authors from 23 different countries. The five countries with the highest number of studies were the USA, Australia, China, the United Kingdom and Spain. Authors linked to the USA contributed 18 different studies (39.1%), authors linked to Australia contributed 8 (17.4%), authors linked to China contributed 7 (15.2%), authors linked to the United Kingdom contributed 6 (13%) and authors linked to Spain contributed 5 (10.9%).
The analysis of the journals in which the studies were published showed that the 46 studies were published in 27 different journals. The top two journals with the highest numbers of publications were the Journal of Travel Research (n=7) and Tourism Management (n=5).

The Data Collection Channels and Sample Sizes of the Studies
The analysis revealed that 52.17% of the studies utilized search engines and databases as data collection channels (n=22). This is followed by browsing in specific journals, with 8.70% (n=4). However, 28.26% of the studies did not specify their data collection channels (n=13). The sample sizes of the studies most frequently were within the range of 26-50 (n=16). Meta-analysis study with the highest sample size examined 222 sample whereas one of the studies did not specify a sample size (Table 3). While the samples of 33 studies consist of publications, the samples of 3 studies consist of case studies. Two of the case studies focused on hotels as research populations and one focused on countries.

The Classification of the Studies According to Research Area Distribution
The analysis determined that the studies were mostly carried out under the research areas of Economic and financial aspects (n=19, 41.3%), Tourist and visitor attitude, value, satisfaction, behaviour and behavioural intention (n=11, 23.9%) and Specific sectors of tourism (health tourism, event tourism, outdoor recreation, leisure, marine, cruises) (n=7, 15.2%) ( Table 4).  1  ----1  2,2  MS&P  ----3  3  6,5  TOUR_SEC  --1  5  1  7  15,2  T&V_Beh  ---1  10  In order to determine which meta-analysis research areas were predominant during which time periods and to reveal current interest towards research areas, the subject titles were examined for specific time periods starting from the year 1994 (Table 4). The analysis showed that no study was carried out on the research areas of Development of theories, models and scales and Methodologies and statistics techniques except in the years 1994-1998. While Tourist and visitor attitude, value, satisfaction, behaviour and behavioural intention and Economic and financial aspects are the most studied research areas, especially in the 2014-2019 period, the research areas of Environmental and cultural issues and Marketing strategies and practices have recently gained attention as well.

The Classification of the Studies According to Research Design
The analysis revealed that 32.61% (n=15) of the studies reflected a correlational construct, 41.30% (n=19) were designed based on regression and 26.09% were carried out experimentally (Table 5). Furthermore, two of the studies that applied meta-regression applied internal meta-analysis, which refers to researchers carrying out meta-analyses of the studies themselves. The analysis of the preferred type of meta-analysis according to the subject titles revealed that most of the studies on the research area of Economic and financial aspects applied meta-regression (n=19), while a great majority of the studies on the research area of Tourist and visitor attitude, value, satisfaction, behaviour and behavioural intention (n=9) had a correlational structure. In addition, the 4 studies published on the research area of Environmental and cultural issues were carried out experimentally (Table 6).

The Analysis of the Meta-Analysis Models Applied in the Studies and the Publication Bias Tests
The analysis of the studies' models determined that 34.78% applied a random-effect model (n=16), 4.35% applied a fixed-effect model (n=2), 10.87% applied both models together (n=3) and the model was not specified in half of the studies (n=23). It was observed that tests for publication bias were applied in 26.09% (n=12) of the studies, whereas 73.91% (n=30) included no analysis of publication bias (Table 7).

Statistics on the Citation Numbers of the Studies
As of June 2019 (the observation date), the most cited study according to the citation data obtained from Google Scholar (GS) and the Web of Science (WoS) database was "Measuring Leisure Motivation: A Meta-Analysis of the Recreation Experience Preference Scales" (n=716), which was published in the Journal of Leisure Research in 1996. This study was carried out in the research area of Development of theories, models and scales with a correlational meta-analysis design. When the citation ratios according to the publication dates of the studies were considered, it was seen that the study with the highest density of citations was "Destination image and tourist loyalty: A meta-analysis", which was published by Zhang et al. in 2012 (Table 8). According to the citation analysis data from Google Scholar, of the 46 total studies, six studies (13.04%) received no citations, nine studies (19.57%) received 1-10 citations, sixteen studies (34.78%) received 11-50 citations, seven studies (15.22%) received 11-50 citations, five studies (10.87%) received 101-400 citations and three studies (6.52%) received over 400 citations.
Analysing the citation numbers based on research area distributions showed that 19 studies carried out on the research area of Economic and financial aspects have the highest citation number (GS=1631). One study (GS=716) carried out on the research area of Development of theories, models and scales and three studies (GS=484) on the research area of Marketing strategies and practices were listed as the most cited research area titles following Economic and financial aspects. Moreover, Development of theories, models and scales (Mean GS=716) and Marketing strategies and practices (Mean GS=161) were the research area titles with the highest citation means according to the publication number (Table 9).

Conclusion
Meta-analysis studies arrived late to the tourism and hospitality literature compared to many disciplines. Long after Glass (1976) put forth the notion of meta-analysis, Crouch (1994) conducted the first meta-analysis study in the tourism and hospitality literature. In the last decade, there has been increasing interest in meta-analysis studies in the field. 82.6% of the studies were published within the decade of 2009-2019. This increase in the number of meta-analysis studies indicates that knowledge of the field is also increasing. In meta-analysis studies, with various methods and data channels (e.g., databases, journals, etc.), the number of samples analysed and the meta-analysis method used should be specified in a hierarchical order. The present paper determined that a great majority of the meta-analysis studies carried out in the field do not specify the data collection channels. For transparency and the reliability of samples, the entire research process, from collecting data to reporting, should be explicated in detail. Within this scope, researchers can apply the data inclusion flowchart and the control lists developed by Moher et al. (2009) called Prisma (Preferred Reporting Items for Systematic Reviews and Meta-Analyses). In addition, most of the studies applied the random-effect model as their meta-analysis model. This shows that the individual studies included in the meta-analysis studies did not have a homogenous structure. A heterogeneous dataset necessitates publication bias tests; however, it was ascertained that the great majority of the studies did not apply publication bias tests. This situation raises questions about whether the individual studies included in the analysis unit have publication bias or not. In order to get an answer to this question, it is necessary to apply to publication bias tests (Funnel plot, Classic failsafe N, Begg and Mazumdar Rank Correlation etc.). However, it is observed that although the random model was used in many meta-analysis studies conducted in the field (that is, despite the heterogeneous structure of the individual studies included in the analysis unit), publication bias tests were not applied. In addition, an important point that stands out is that half of the studies did not report the model they applied. This situation creates suspicion that the study is based on its reliability, transparency and ethical principles. Meta-analysis studies in the tourism and hospitality literature can direct theory development, shed light on conflicting findings and lead to improved methodologies (Gretzel &Kennedy-Eden, 2012, p.466). For this reason, meta-analysis studies should be conducted in line with the principles of transparency and reliability, starting from the data collection stage and continuing to the reporting stage. Ensuring consistency is important for obtaining reliable findings in the tourism and hospitality field, as this field encompasses many different disciplines (Pahlevan-Sharif, Mura & Wijesinghe, 2019, p.164-165).
This paper showed that meta-analysis studies in the field are mostly conducted on the economy, tourist perceptions and behaviour and alternative tourism types. Although the first meta-analysis study in the field were conducted on tourism demand and demand elasticities, the majority of meta-analysis studies were carried out on the economy and tourist perceptions, attitudes and behaviour (satisfaction, loyalty), especially in the last five years. Kim et al. (2018) also ascertained that systematic review studies conducted within the tourism and hospitality discipline intensively focus on the research areas of economy and finance. This finding can be a solution to the various challenges faced by the industry, such as economic instability, stagnancy and the distribution of tourism gains. When meta-analysis is considered as a combination of multiple studies, the fact that there are so many meta-analyses conducted on these research areas shows the high number of individual studies on these research areas as well. For instance, the industry's inclination towards alternative tourism types, after the saturation of mass tourism, is reflected in research and the numbers of researches carried out on this area has increased.
When the studies were evaluated according to research design, it was observed that regression-based designs are more frequent. Especially in recent periods, there is increasing interest in meta-regression studies. Regression-based studies have mostly focused on economic research area matters. Correlational meta-analysis is frequently observed in studies examining tourist perceptions and behaviour. In contrast, studies focusing on environmental and cultural values have an experimental design. Additionally, structural equation modelling (SEM) is becoming an increasingly popular statistical technique among tourism studies where it is used to test various types of theoretical models (Reisinger & Turner, 1999;Nunkoo & Ramkissoon, 2012;Nunkoo et al., 2013). However, meta-analysis studies using SEM results are very limited in the literature. Future meta-analysis studies, taking into account Hedges' (2016) views, can conduct meta-analysis studies based on SEM results and make significant contributions to this gap in the literature.
The results of the citation analysis reveal that the most cited meta-analysis studies were conducted on the research areas of theory, model and scale development, marketing strategies and applications and the economy. For example, although there were only a few studies conducted on the research areas of theory, model and scale development and marketing strategies and applications, they received much attention in terms of citation numbers and citation densities. This combination of a few studies with high impacts can be a beacon for future studies.
With an exploratory perspective, the paper serves researchers as a reference for the scope, limitations, tendencies and future effects of the meta-analysis studies conducted within the field. This paper identifies the current status (historical development, scope, limitations, trends) of the meta-analysis studies carried out in the field of tourism and hospitality and provides inferences and suggestions for its future effects. The findings obtained through the systematic review of the meta-analysis studies contribute to a better understanding of the meta-analysis studies carried out in the field but also provide information about how future research should be structured. In this context, in the current paper, provided information about the historical development of meta-analysis studies in the field, and by identifying the authors, countries and journals leading the meta-analysis studies, a resource for future research was provided. In addition, by giving information about data collection channels of metaanalysis studies, sample numbers, applied meta-analysis models and publication bias tests, important and necessary criteria in a meta-analysis study and basic information about what these criteria mean are presented. This information provides critical clues to future research on how a meta-analysis study should be structured. In addition, classifying meta-analysis studies according to research area distributions and research designs will allow future researchers to identify trends, concentrations and gaps in the field. Finally, determining the meta-analysis studies that are effective in the literature based on the research area and research design through citation analysis, will undoubtedly be a guide for future researchers in the selection of research area and research design. In this respect, the paper will serve as a starting point and checklist for future meta-analysis researches to be carried out in the field of tourism and hospitality, provide researchers with insight into potential future research opportunities, and help them explore research topics. Therefore, it is thought that the paper will contribute to the advancement of scientific development in the field by putting forward an idea about the current state of meta-analysis studies in the field and making determinations and inferences that guide the future meta-analysis studies.
This paper has certain limitations. Although a comprehensive literature review was carried out during the data collection stage, a limitation is that some studies were in different languages, had different scientific genres (proceedings, book chapter, thesis, etc.) and had inaccessible full-text. Searches could not be carried out in the tourism and hospitality journals in the Web of Science included within the scope of ESCI; however, it is believed that the other databases and the detailed examination of Google Scholar resolved this limitation.