Meta-analysis involves secondary data gained from a number of studies which have already been conducted on the same topic. When there is conflicting evidence from numerous studies on a particular topic, psychologists may conduct a meta-analysis, or an analysis of existing analyses. They will use a database to search for studies which fit certain criteria such as studies on the effectiveness of CBT with service-users who have recently been released from residential care.
Example of meta-analysis: In 2009, The National Institute for Health and Care Excellence (NICE, 2009) published a meta-analysis looking at cognitive therapy for schizophrenia demonstrating that this psychological treatment is “effective in reducing rates of readmission to hospital and duration of admission. It was also judged to be effective in reducing overall symptom severity, both at the end of treatment and after up to 12 months’ follow-up. Effectiveness against positive symptoms was more limited, but evidence was marshaled for benefits on hallucinations” (McKenna, 2014). The analysis included studies of cognitive-behavioural therapy versus standard care and cognitive-behavioural therapy versus other active (psychological) treatments.
McKenna goes onto explain a variety of flaws in this meta-analysis however, and these flaws are often the case in other studies too.
- It examined a large number of outcome measures and clumped them altogether under as positive outcomes, however, there is a great deal of diversity with regard to the measures that were used e.g.
- relapse and readmission rates
- social and occupational functioning
- mortality and suicide
- many of the measures were non-independent, e.g. they counted symptoms experienced by the same person at at six, 12, and 18 months’ follow-ups
- they summed positive symptoms as well as separate ratings for delusions and hallucinations, the latter with various subsidiary measures such as command hallucinations, malevolence, and omniscience
- Ability to reveal trends in conflicting sets of data
- Large data sets due to multiple studies being used means the conclusions should be more generalisable, especially if studies have been conducted in different countries for example
- Often the studies that are included do not have the same procedures or use different measures meaning that like is not being compared with like, rendering the results, potentially, rather meaningless
- The meta-analysis is only as useful as the studies that is it made up from; if the studies have issues with validity or reliability for example, then the meta-analysis will not be useful either
- Studies which lead to non-significant findings are often not published, leading to a publication bias and thus false impressions about certain topic areas; this means that meta-analyses may also be biased if they do not have access to all the data on a certain topic as much of it is not published
Another example of meta-analysis: The gene database on the Schizophrenia Research Forum (http://www.schizophreniaforum.org) provides a systematic and regularly updated meta-analysis of genetic association studies.