Back to Chapter Back to Chapter

Case 2g :

Falsification—If at first the data doesn't fit, try, try to make it fit

"Social science is the slow boring of hard boards". Zeynep tried to remember this phrase from Max Weber, famed German sociologist, when she ran into problems with analysis of the data she gathered for her PhD thesis.

Zeynep spent two years trekking through the jungles of Papua New Guinea, the Philippines, and French Guyana, speaking to community leaders about their experience with international pharmaceutical firms engaged in "bioprospecting" on their native lands. She had recorded and transcribed hundreds of hours of interviews, thousands of photographs, and page upon page of maps.

The trouble was, the data in the interviews, in particular, pointed away from her hypothesis: pharmaceutical firms were engaging local leaders intensely and often. This contradicted almost all of the literature. In fact, when Zeynep told her supervisor of this surprising finding, her supervisor insisted that Zeynep must have miscoded some of the conversations and asked that she re-review the transcripts. Having completed a re-examination of her coding scheme and a full re-code of all of her transcripts, plus re-running the analyses twice, Zeynep now has stronger evidence in this data that engagement is the norm, not the exception.

But, now she has a problem. Her supervisor is a major contributor to the debate that suggests that pharmaceutical firms engage in "rapacious bioprospecting". Her supervisors' academic career and many public appearances area based upon her supporting this contention. Now, one of her students has data showing this may not be the case in some sensitive areas!

Zeynep feels she is in a quandary: she is certain that her analyses are accurate, but she is also certain that she will have difficulty gaining her supervisors' support of her final research outcome. Zeynep thinks,

"Certainly my supervisor must be correct. Maybe there were problems with translation. I can explain the apparent contradiction as a problem of language, at least. I could also reverse the coding scheme for frequency of interaction and intensity of interaction… this would likely reduce the effect of intensity of interaction on the analysis.",

Laying awake in bed at 3am, Zeynep is torn:

"Do I follow my data or do I follow my supervisor"?

• What is the core problem here?
• What can you identify as Zeynep's biggest concern?
   

Imagine you are Zeynep's best friend, with whom she shares the problem over a croissant and coffee this morning:

• What would you tell Zeynep to do?
• Should Zeynep change her research to follow her supervisor's research conclusions?
• Should she follow her own data?
• What if she follows the trail in her data and fails to win the support of her supervisor?
• Should Zeynep seek advice before or after submitting any data, and if so, where might Zeynep go for advice on this matter?