Pros
The salary and benefits are very good compared to other policy research organizations. Mathematica has a rigorous internal process for producing good deliverables, although these products lack flare. Staff are smart and capable. You have the opportunity to work in different policy areas from labor to early childhood. Work life balance is respected.
Cons
The organization lacks commitment to diversity beyond it being a "buzz word" despite diversity awards, community service work, and a nicely written but hollow diversity statement on the corporate website. There is no diversity/cultural competency training, no mentorship or affinity group programs, diverse employees are not asked to help with recruitment, and diversity/cultural competence is not a topic in interviews. Mathematica does have a summer fellow program for people doing research on minority groups but ironically few of these fellows are racial and ethnic minorities. There are few people from underrepresented racial and ethnic groups in the organization, especially in the research divisions and among senior management. Staff exhibit very little cultural competency or awareness and can be very quick to judge or assume without trying to understand something from a different cultural or personal perspective. However, there is a lot of gender diversity and limited diversity of Asian American/Pacific Islander, Arab American, and international backgrounds. I admit that I am not an underrepresented minority but if you are and are considering Mathematica you should find an organization that does more to actually support you professionally day to day and holds itself and employees accountable for diversity. Professional development opportunities are limited: there's no professional development plan and few company sponsored professional development opportunities. Your development as a research professional is not going to be about learning new theories in your discipline, research and evaluation design methods, or new quantitative or qualitative tools and techniques but learning about how Mathematica does these things. This is an important distinction since Mathematica and its senior managers are not necessarily up on the most current research or practices and many are unwilling to use them since they are not safe or familiar. The culture is very uniform. The structure is very hierarchical (which some may like). These two things stunt innovation, growth, and change. In general, the company goes with what's been done before because it has worked in the past rather than try to develop new or improved practices. Therefore, reports and briefs are a cookie cutter version of one another regardless of whether you are studying an early childhood program or an employment program. Also, internal policies do not change and suggestions may be encouraged but will probably not result in anything actually happening. Success depends on your willingness to do things the "Mathematica way" and conform not your capacity to truly lead by developing new methods. Managers and staff are nice but not necessarily open minded, emotionally intelligent, or supportive and can be very judgmental toward things that are different. The company does a poor job of integrating new staff and getting them good work; it may take 6 months to 2.5 years for you to truly begin to get work that will interest you. This may have been fine in the past but this is an era where workers are expected to change jobs every few years, especially if their work is unsatisfying. Also, because the culture and people are so uniform, you'll have a hard time if you come from a background or have experiences and perspectives that are different from other staff. I would say this is especially true in research.