Positive Deviant Validation Evaluate the performance of positive deviants over time and check whether they consistently outperform their peers The timeframe will vary according to the focus of the project and selecting it can require expert opinion Time series analysis Identifying potential positive deviants across rice producing areas in Indonesia https www unglobalpulse org document identifying potential positive deviants pds across rice producing areas in indonesia an application of big data analytics and approaches The last step of this stage consists of initial validation of the potential positive deviants you identified While field research is necessary for full confirmation of their deviance there are intermediate ways to validate whether observed deviance in the data is simply due to random noise or whether it is based on actual signals of outperformance Here are several ways to do so In the Indonesia agricultural project the team conducted a time series analysis to see whether the performance of rice farming villages was independent of climatic patterns They developed a model to predict the EVI at the village level as a function of precipitation and temperature in 2013 where they used historic climate and EVI data from 2000 to 2012 to train the model The observed performance of positively deviant villages was significantly higher than the observed performance of non positively deviant villages This implied that outlier villages had likely adopted specific approaches and practices that others had not and had established production systems that delinked climatic patterns and productivity This provided an initial validation of them being potential positive deviants Local knowledge Reach out to community leaders government officials local domain experts and development practitioners who are engaged in activities projects or services related to the targeted areas before doing the field research Sharing with them the initial set of potential positive deviants can lead to possible explanations for outperformance that may have been overlooked and could bias the results without reflecting the actual practices you are interested in Explanations such as the existence of development interventions in positively deviant areas i e external support or having better access to resources like irrigation points Check whether possible predictors of PD can be identified through official statistical data like a census or survey or other sources of data such as high resolution imagery covering the targeted areas Although the PD approach is mainly concerned with identifying grassroots practices which are seldom captured through such data some higher level management practices can be distinguished If their association with positively deviant units is in accordance with existing literature and local knowledge this can count as means of validation in itself Triangulating different data sources 67 68

Vorschau DPPD Handbook Seite 35
Hinweis: Dies ist eine maschinenlesbare No-Flash Ansicht.
Klicken Sie hier um zur Online-Version zu gelangen.