SSudo's Lab



書誌情報:Parkhusrt, J. (2017). The politics of evidence (R)

The political perscpective used in this book starts from a recgnition that policymaking is typically concerned with setting priorities and allocating scarce resources. In doing so, policy decisions typically involve choices between options containing multiple and competing sets of social values. From this perspective, when presented with evidence that something works, the natural response should not be to simply do it, but rather to ask: 'Works to do what, exactly?' In other words, presenting evidence that something is effective does not necessarily mean that it is socially important. (p. 19, 下線は引用者)


Medical interventions are based around biological and psysiological mechanisms which are widely shared by humans. Many other interventions (like those to reduce crime or promote better educational outcomes) will function through socially embedded mechanisms that may not be common or that at least need some additional information to assume commonality across contexts. Medicine provides a great inspiration, but the human body is fundamentally different from a social enviroment. (p. 21, 下線は引用者)

人体の構造には国や地域に関係なくある程度の同質性 (commonality) があるため、例えば、アメリカで実施された薬学の成果がアメリカに特有のものであって、それ以外の国では当てはまらないという事例は起きにくい。一方、犯罪の抑止や教育効果を高めるための施策は、国・地域が違えば「環境」が異なるため、当該地域の研究成果の一般化は慎重に行われる必要がある。このような「外的妥当性」の重要性が顕著である点が、薬学と政治学の大きな違いの内の一つであろう。

Nutley et al. explain that it is policy makers and practioners' who use research in strategic and technical ways, noting that: 'Policy makers say that while research is often interesting and helpful... it most often "informs" policy, rather than providing a clear steer for action' (2007, p. 37).
Despite this fairly extensive body of work mapping out the multiple ways in which research can be utilised in policy process, the EBP literature still overwhelmingly reflects the idea that evidence use is a techinical problem-solving exercise (Greenhalgh and Russell 2009). However, this focus on problem solving shows its limitations quite quickly when considering how few policy decisions actually fit this model. (p. 25, 下線は引用者)


The main issue is not that RCTs and hierarchies are inherently flawed, but rather that they are being incorrectly applied in many cases if they are used to prioritise policy choices. As such, appeals to hierarchies can impose issue bias if they result in prioritising those social concerns conductive to experimentation or where stakeholders have already conducted experiments (Barnes and Parkhurst 2014). Over-reliance on hierarchies can also obscure the importance of external validity, often failing to explicitly address questions of the applicability of findings across contexts. (...) There still needs to be critical reflection upon what hierarchies can be used for and what 'good evidence for policy' whould have to look like if single hierarchies do not meet the needs for evidence use within policy decisions. (p. 29)

"good evidence" と "good evidence for policy" は異なる。前者はエビデンス階層や外的妥当性・内的妥当性などの基準が担保されていることのみを求めるのに対し、後者はそれに加え、社会的需要との一致、実施の容易性などの社会的要素・状況を考慮する必要がある。その意味で、EBP とは「ネゴシエーション」が決定的に重要であるといえる。この点を考慮すると、政策研究の論文でのインプリケーションは、けっこう慎重に書かないとダメな気がする。単に、「現状はAを採用しているが、AよりもBの方が効果が高いことがわかった。よって、Aを廃止しBを採用すべきだ」との論調で政策提案をしてしまうことは不適切。そうではなく、そこは研究者の立場として、「AよりもBの方が効果が高いことがわかった」という結果の提示のみに留め、それをどのように活用するかは行政担当者に委ねる。あるいは、どうしても政策的な提言をしたいのであれば、コストパフォーマンスの問題や、その施策の実現可能性を丁寧に記述したうえで、控えめに政策的示唆を付け加えるのがベストだと思う。

While the above cases illustrate how the research process can be manipulated to create biased evidence, techinical bias can also occur in the selection of evidence, when a body of (potentially technically valid) evidence is cherry-picked so as to only highlight those pieces of evidence which support a desired outcome. This is particularly a proble in policy debates touching on complex or uncertain issues, as in such situations there can be many pieces of relevant information, and such information may be contradictory. Indeed, it is very rare to have all evedence and all studies showing the same outcomes or the same direction of effect in any scientific field of enquiry. Often there is a range of findings, and it is necessary to look at the totality of the evidence to discern a pattern or overall trend. This is frequently why systematic reviews of research are so important in the EBP field, as they aim to follow explicit steps to ensure that all relavant evidence is considered use of pieces of evidence allows groups to focus on different facts in line with their political needs and goals. (p. 47, 下線は引用者)

社会科学の研究では、同じ手続きを用いて実験・介入を行ったとしても、先行研究と同じ結果になることはむしろ珍しい。だからこそ、複数の研究を統合する systematic review が重要である——との指摘。自分の都合に合わせてエビデンスをつまみ食いするのではなく、自分の仮説に合わない研究についても、それが内的妥当性・外敵妥当性の担保されているものであれば、無視してはならない。というより、そういった「外れ値」にある研究結果と向き合うことで、自分の仮説をさらに鍛え上げることができる。