This study examined the statistical consequences of employing various methods of computing and cumulating effect sizes in meta-analysis. Six methods of computing effect size, and three techniques for combining study outcomes, were compared. Effect size metrics were calculated with one-group and pooled standardizing denominators, corrected for bias and for unreliability of measurement, and weighted by sample size and by sample variance. Cumulating techniques employed as units of analysis the effect size, the study, and an average study effect. In order to determine whether outcomes might vary with the size of the meta-analysis, mean effect sizes were also compared for two smaller subsets of studies. An existing meta-analysis of 60 studies examining the effectiveness of computer-based instruction was used as a data base for this investigation. Recomputation of the original study data under the six different effect size formulas showed no significant difference among the metrics. Maintaining the independence of the data by using only one effect size per study, whether a single or averaged effect, produced a higher mean effect size than averaging all effect sizes together, although the difference did not reach statistical significance. The sampling distribution of effect size means approached that of the population of 60 studies for subsets consisting of 40 studies, but not for subsets of 20 studies. Results of this study indicated that the researcher may choose any of the methods for effect size calculation or cumulation without fear of biasing the outcome of the metaanalysis. If weighted effect sizes are to be used, care must be taken to avoid giving undue influence to studies which may have large sample sizes, but not necessarily be the most meaningful, theoretically representative, or elegantly designed. It is important for the researcher to locate all relevant studies on the topic under investigation, since selective or even random sampling may bias the results of small meta-analyses.

The burden of depression requires modalities other than individual psychotherapy if we are to reduce it. Over the past two decades preventative programs for depression have been developed and refined for different populations. The six years since the last meta-analysis of preventative interventions—inclusive of all program types—have seen a number of new studies. The current study used the greater statistical power provided by these new studies to analyze moderators of, and sub-group differences in, the effect of these interventions on depression. Moreover, this meta-analysis synthesized effect sizes for outcomes other than, but often related to, depression (e.g., anxiety) and for within-group change scores with the goal of better informing program implementation and evaluation. Twenty-nine studies met inclusion criteria and indicated that small, robust effects exist for reductions in depression diagnoses and symptomatology. Significant effects were also observed for anxiety, general health, and social functioning.

Pork quality is an important issue for the whole meat chain, from producers, abattoirs, retailers through to costumers and is affected by a web of multi-factorial actions that occur throughout the pork production chain. A vast amount of information is available on how these diverse factors influence different pork quality traits. However, results derived from individual studies often vary and are in some cases even contradictory due to different experimental designs or different pork quality assessment techniques or protocols. Also, individual influencing factors are often studied in isolation, ignoring interacting effects. A suitable method is therefore required to account for a range of interacting factors, to combine the results from different experiments and to derive generic response-laws. The aim of this thesis was to use meta-analyses to produce quantitative, predictive models that describe how diverse factors affect pork quality over a range of experimental conditions.

The purpose of this study was to quantitatively evaluate the overall effectiveness of eating disorder prevention programs and to investigate potential moderating variables that may influence the magnitude of intervention effects. Meta-analysis was used to conduct a comprehensive and systematic analysis of data across 46 studies. Effect size estimates were grouped into outcome sets based on the following variables: knowledge, general eating pathology, dieting, thin-ideal internalization, body dissatisfaction, negative affect, and self-esteem. Q statistics were used to analyze the distribution of effect size estimates within each outcome set and to explore the systematic influence of moderating variables. Results revealed large effects on the acquisition of knowledge and small net effects on reducing maladaptive eating attitudes and behaviors at posttest and follow-up. These programs were not found to produce significant effects on negative affect, and there were inconsistent effects on self-esteem across studies. Population targeted was the sole moderator that could account for variability in effect size distributions. There was a tendency toward greater benefits for studies targeting participants considered to be at a relatively higher risk for developing an eating disorder. Previous assumptions regarding the insufficiency of "one-shot" interventions and concerns about the iatrogenic effects of including information about eating disorders in an intervention were not supported by the data. These findings challenge negative conclusions drawn in previous review articles regarding the inability of eating disorder prevention programs to demonstrate behavioral improvements. Although these findings have implications for the prevention of eating disorders, it was argued that a clear link between intervention efficacy and a decreased incidence of eating disorders was not demonstrated. Rather, only direct information was offered about the ability to influence eating disorder related knowledge, attitudes, and behaviors. Specific recommendations related to intervention content, reasonable goals/expectations, and outcome criteria were offered for improving research in this area.

Functional Communication Training (FCT) is an intervention that involves teaching a communicative response to decrease the occurrence of challenging behavior in individuals with disabilities. FCT is a two step intervention in which the interventionist first determines the function, or purpose, of the challenging behavior and then teaches a communicative response that will provide the same function as the challenging behavior. This meta-analysis addressed the following questions: (a) Is FCT more effective with a complete or brief functional analysis? (b) Is FCT differentially more effective for one communication mode versus another (unaided augmentative and alternative communication, aided augmentative and alternative communication, or verbal)? (c) Is FCT more effective when implemented in natural or contrived contexts? (d) Is FCT more effective for different functions of challenging behavior (attention, tangible, escape and multiple)? (e) How effective is FCT with individuals with challenging behavior, across different age ranges? (f) How effective is FCT with individuals with challenging behavior, across different disability categories? A thorough search was performed to find all articles related to FCT. The articles were then reviewed to ensure that they met the inclusion criteria. Data were extracted from the graphs within each study and then analyzed using Robust Improvement Rate Difference (IRD). Forest plots were also created to aid in visual analysis to determine statistical significance and consistency of the results. A variable was determined to moderate the effectiveness of FCT if there was a statistically significant difference between the levels within each variable. Thirty nine studies were included in this meta-analysis. Over-all FCT has a Robust IRD score of .86 (confidence intervals = .85 - .87). Based on the findings of this meta-analysis FCT is most effective with brief functional analysis and verbal communication. FCT was equally effective in natural and contrived settings. FCT appears to be most effective when an individual's behavior serves as attention seeking or an attempt to gain access to a tangible item. FCT appears to be more effective with school age individuals rather than adults. Finally, FCT may be more effective with individuals with autism spectrum disorder than intellectual disabilities or other disabilities.

Species interactions restrict or promote population growth, structure communities, and contribute to evolution of diverse taxa. I seek to understand how multiple species interactions are maintained, how human altered species interactions influence evolution, and explore factors that contribute to variation in species interactions. In Chapter 1, I examine how plants interact with multiple guilds of mutualists, many of which are costly interactions. The evolution of traits used to attract different mutualist guilds may be constrained due to ecological or genetic mechanisms. I asked if two sets of plant traits that mediate interactions with two guilds of mutualists, pollinators and ant bodyguards, were positively or negatively correlated across 36 species of Gossypium (cotton). Traits to attract pollinators were positively correlated with traits to attract ant bodyguards. Rather than interaction with one mutualist guild limiting interactions with another mutualist guild, traits have evolved to increase attraction of multiple mutualist guilds simultaneously. In Chapters 2 and 3, motivated by the fact that agriculture covers nearly 50% of the global vegetated land surface, I explore the consequences of changes in plant mutualist and antagonist guilds in agriculture for selection on plant traits. I first explore how agriculture alters abundance and community structure of mutualist pollinators and antagonist seed predators of wild Helianthus annuus texanus. Mutualists were more abundant near crops, whereas antagonists were more abundant far from crops near natural habitat. In addition, mutualist pollinator communities were more diverse near sunflower crops. Plant mutualists and antagonists respond differently to agriculture. Next, I explore how these changes in abundance and community structure of mutualists and antagonists influenced natural selection on H. a. texanus floral traits. Natural selection on heritable floral traits differed near versus far from crop sunflowers, and overall selection was more heterogeneous near crop sunflowers. Furthermore, mutualist pollinators and antagonist seed predators mediated these differences in selection. Finally, in Chapter 4, I ask if variation in interaction outcomes differs across types of species interactions. Furthermore, I examined the relative importance of factors that create context-dependency in species interactions. Using meta-analysis of 353 papers, we found that mutualisms were more likely to change sign of the interaction outcome when compared across contexts than competition, and predation was the least likely to change sign. Overall, species identity caused the greatest variation in interaction outcomes: whom you interact with is more important for context-dependency than where or when the interaction occurs. Additionally, the most important factors driving context-dependency differed significantly among species interaction types. Altogether, my work makes progress in understanding how species maintain interactions with multiple guilds of mutualists, how agriculture alters species interactions and subsequent natural selection, and the variation in species interaction outcomes and their causes.

Many fields of economics point to technology as the primary vehicle for change. Agencies pushing change often promote technology adoption to achieve their goals. To improve our understanding of how efforts to push new technologies should be focused, two studies are undertaken. The first study defines and tests for universality using meta-regression analysis on 170 analyses of agricultural production technologies. The second study, a case study on an emerging information technology - climate forecasts, examines how the timing of adoption affects the benefits. A factor exhibiting a systematic positive or negative effect on technology adoption is a universal factor. If the impact is the same regardless of location or technology type, the factor is strongly universal. The factor is weakly universal if the impact varies by location or technology type. Education and farm size are found to be weakly positive universal, age is found to be weakly negative universal, and outreach is not found to be a universal factor in the adoption of technology. These results indicate that technology-promoters may want to change their approach and focus on younger, more educated producers with larger farms. In the second study, an international wheat trade model incorporating climate variability is used to simulate different scenarios when wheat producers in the U.S., Canada, and Australia adopt ENSO-based forecasts for use in production decisions. Adoption timing and levels are varied across countries in the different scenarios. The results are highly consistent. Early adopters benefit the most, there is no incentive for more producers to adopt after 60% to 95% have adopted (meaning the adoption ceiling has been reached), and slower adoption corresponds to ceilings closer to 60% than 95%. Examining technology adoption from two angles provides a deeper understanding of the adoption process and aids technology-promoters in achieving their goals. In addition to focusing on younger, more educated producers with larger farms, technology-promoters wanting wide-spread adoption with high benefits need to push constituents to adopt early and fast.

The purpose of this study was to quantitatively evaluate the overall effectiveness of eating disorder prevention programs and to investigate potential moderating variables that may influence the magnitude of intervention effects. Meta-analysis was used to conduct a comprehensive and systematic analysis of data across 46 studies. Effect size estimates were grouped into outcome sets based on the following variables: knowledge, general eating pathology, dieting, thin-ideal internalization, body dissatisfaction, negative affect, and self-esteem. Q statistics were used to analyze the distribution of effect size estimates within each outcome set and to explore the systematic influence of moderating variables. Results revealed large effects on the acquisition of knowledge and small net effects on reducing maladaptive eating attitudes and behaviors at posttest and follow-up. These programs were not found to produce significant effects on negative affect, and there were inconsistent effects on self-esteem across studies. Population targeted was the sole moderator that could account for variability in effect size distributions. There was a tendency toward greater benefits for studies targeting participants considered to be at a relatively higher risk for developing an eating disorder. Previous assumptions regarding the insufficiency of "one-shot" interventions and concerns about the iatrogenic effects of including information about eating disorders in an intervention were not supported by the data. These findings challenge negative conclusions drawn in previous review articles regarding the inability of eating disorder prevention programs to demonstrate behavioral improvements. Although these findings have implications for the prevention of eating disorders, it was argued that a clear link between intervention efficacy and a decreased incidence of eating disorders was not demonstrated. Rather, only direct information was offered about the ability to influence eating disorder related knowledge, attitudes, and behaviors. Specific recommendations related to intervention content, reasonable goals/expectations, and outcome criteria were offered for improving research in this area.

Thesis (doctoral)--Universität, Hamburg. / Includes bibliographical references and index.

published_or_final_version / Medical Sciences / Master / Master of Medical Sciences