External stresses mediate an association between negative perception of activism and a decision never to participate

Thomas Levine and Mimiko Watanabe

Research question

We wanted to test whether a desire to participate in activism was associated with increased participation in activism and whether this relationship was mediated by external stresses.

+-----------------+                   +-----------------+
| Interest in     |                   | Participation   |
| participating   |   ------------>   | in activism     |
| in activism     |                   |                 |
+-----------------+        / \        +-----------------+

                   | External stress |
                   | (Barriers to    |
                   | participation)  |

When we discuss “interest in participation”, we are referring to more fundamental drivers in considering whether to participate, such as agreement with the goals of the activism or belief that the activism will be effective.

“External stresses” are secondary, exogenous considerations that might prevent participants from participating in activism that they would support in concept. For example, one strong external stress would be working multiple jobs.

The questionnaire

A questionnaire was developed using critical participatory action research* as the methodological approach. It was based on the collective experiences of the researchers as CUNY students. Several pilot surveys were distributed on CUNY campuses and at the Free University, an activist event in Madison Square Park that occurred on May 1, 2012. This enabled researchers to revise and build the survey based on feedback from students. Insight from CUNY students was essential in the construction of the survey. Distribution of the final version of the survey began May 1, 2012 and continued until September 16, 2012. Surveys were collected on CUNY campuses, in classrooms, at CUNY activist events, and online.

* Torre, M. E. and Fine, M. (2011), A Wrinkle in Time: Tracing a Legacy of Public Science through Community Self-Surveys and Participatory Action Research. Journal of Social Issues, 67: 106–121. doi: 10.1111/j.1540-4560.2010.01686.x

Questionnaire distribution

Researchers made contact on various CUNY campuses with professors who might be willing to distribute surveys to students during class time. An effort was made to ensure that the various types of CUNY campuses were included in the study (i.e. community colleges, senior colleges, etc) as well as various types of classes (day classes, night classes, and summer classes). If professors responded positively, a time was arranged during class when researchers could distribute surveys.

Surveys were distributed to children’s studies classes at Brooklyn College, sociology classes at Hunter College, psychology classes at LaGuardia, research methods classes at the CUNY Graduate Center, classes at Medgar Evers college, and classes at the School of Professional Studies. A researcher was present in each classroom to oversee the completion of surveys and collect them. Researchers gave participants a short description of the survey, indicating that researchers were CUNY students interested in studying activism on CUNY campuses. Anonymity of participants was assured. Consent from participants was obtained verbally and they were given about ten minutes to complete the survey. One professor provided students with a link to fill out the survey online.

The remaining surveys were distributed during activist events, including the Free University at Madison Square Park on May 1st, 2012. The Free University was an event where workshops and classes were set up by student activists in a public space (Madison Square Park) and open to the public. Many CUNY students attended the event, and some took the time to fill out the survey. Surveys were also collected at the CUNY Graduate Center general assembly, which is a student activist event where students meet to discuss issues and organize. Surveys were also distributed in the Hunter College cafeteria and student lounge, as well as at a student speak-out activist event to students who were willing to complete them. In these cases, researchers gave participants ten minutes to complete the survey and acquired verbal consent.

When researchers obtained a substantial number of surveys from a diverse group of CUNY campuses, data collection ceased. A code book for the survey was created, and all researchers participated in data entry, and eventually compiled all entered data into one Excel spreadsheet.

Representativeness of the questionnaire

This sample consisted of 304 current students attending various CUNY campuses. The survey was distributed on a total of twelve CUNY campuses, mostly from Hunter College (29% of sample), Brooklyn College (19%), Medgar Evers College (16%), Laguardia Community College (13%), and the Graduate Center (10%). The remaining CUNY schools include the School of Professional Studies (5%), Baruch College (1.6%), John Jay College of Criminal Justice (.7%), Borough of Manhattan Community College (.3%), City College of New York (.3%), Lehman College (.3%), and Queens College (.3%). Several participants did not provide their campus name (5%). Students were pursuing various degrees, most commonly a Bachelors degree (60%), followed by Associates (13%), Doctorate (10%), Masters (5%), Certificate (1%), or Non-degree seeking students (1%), and 11% were unknown.

When asked for their gender, the majority of participants identified as female (67.4%), while 27% were male, 2% were queer, and 3% did not include their answer in this portion of the survey. Official records state that CUNY students are 58% female and 42% male (CUNY Office of Institutional Research and Assessment, 2011), although it should be noted they do not include alternative gender categories. It also states that 28.2% of CUNY students are age 25 or older. The ages of this sample ranged from 18 to 58, with 50% of participants between the ages of 21 and 29. The average age of this sample was 26.5.

Based on all this, the sample seems fairly representative of the CUNY student body.


Preferably, we would have had each participant answer the same questions about her level of interest of participation (such as whether she agrees with the movement’s views or whether she think the movement will have positive impact) and her external stresses (like working overtime) and her level of participation. The questionnaire wasn’t quite like this, so we had to adapt the questions somewhat.

The questionnaire asked one question about level of participation to all participants (question 8).

8 Have you ever attended an event on Occupy CUNY/ activism for higher education (e.g., a march, a talk, a general assembly/GA)?

  • Yes, and I want to participate again because ________
  • Yes, but I do not want to participate again because ________
  • No, and I will not participate because ________
  • No. I would like to participate but ________

The questions about interest in participation (question 7) and about external stress (question 4) were arranged such that people answered different questions depending on whether they had previously participated in OccupyCUNY, which is quite related to their level of participation in activism.

3 Are you participating in activism that addresses issues in higher education?

  • Yes (go to 4a and 4b)
  • No (go to 4c)

4a If Yes, what motivates you? ________

4b If yes, what enables you to participate?

  • Financial resources
  • Job security
  • Childcare
  • Sense of belonging/community
  • Training in activism skills
  • Having a flexible schedule
  • Support from family, friends, peers, partners
  • Encouragement at work and union
  • Access to ongoing sources of info
  • Encouragement at school (e.g., Class credits, teacher support, teacher allows me to miss class, inclusion in my academic research)
  • Other concrete things like _______

4c If no, what prevents you from participating?

  • Lack of financial resources
  • Job insecurity
  • No childcare
  • Inflexible schedule/No time
  • No support from family, friends, peers, partners
  • Institutional discouragement/limitations at work and union
  • No access to ongoing sources of info
  • Institutional discouragement/limitations at school
  • Surveillance
  • Police violence/threat
  • Consequence of arrests
  • Other _______

Question 4 is above, and question 7 is below.

7a Overall, how do you feel about Occupy CUNY/ activism for higher education? Check only one.

  • Positive
  • Negative
  • Ambivalent/ mixed feelings
  • I don’t know much about it
  • I don’t have an opinion

7b I feel positive about Occupy CUNY/ activism for higher education because: (Check all that apply.)

  • It opens a dialog about privatization of public higher education
  • It is led by students for students
  • I feel a part of a movement
  • It represents common issues faced by the majority of the population
  • It uses a democratic process
  • It uses nonviolent methods
  • It uses direct action
  • Other (please specify) ________

7c I feel negative about Occupy CUNY/ activism for higher education because: (Check all that apply.)

  • It is not effective
  • It does not deal with issues that concern me
  • It is unfocused, and does not have a coherent message
  • It is disruptive to classes
  • Other (please specify) ________

We thus had to separate the analysis by whether people had previously participated in OccupyCUNY.

There wasn’t much separation by level of participation within the people who said they had participated. (There wer 64 people in the higher group and 7 in the lower group.) Thus, we only conducted the analysis for the respondents who said they had not participated in OccupyCUNY. This led to a reversed framing of the relationship described above.

+-----------------+                   +-----------------+
| Negative atti-  |                   | Decisiveness    |
| tude towards    |   ------------>   | about not       |
| activism        |                   | participating   |
+-----------------+        / \        +-----------------+
  (question 7c)             |            (question 8)

                   | External stress |
                   | (Barriers to    |
                   | participation)  |
                      (question 4c)


The questionnaire requested that question 4c only be answered by people who had answered with one of the two “No” responses to question 8, so we used only the subset of the data for which question 8 was one of these no responses. We thus had a binary participation variable, no.never, which was true if the participant said she would never participate and false if she said that she had not participated but would like to.

For both questions 4c and 7c, almost people checked either zero or one box. For simplicity, we simply checked whether at least one of the respective groups of check marks was checked. Thus, our external stress variable, has.external.stress, was true if at least one of question 4c’s boxes was checked and false if none was checked. Similarly, our variable for negative perception of activism, has.negative.perception, was true if at least one of 7c’s boxes was checked and false if none were.

We chose variable names that start with the word “has” so that it would be less confusing to talk about, but both groups obviously have both external stresses and some negative perception of activism; if these names bother you, mentally switch the “has” to “has more”.


For our sample, here are the proportions of people within each of the four groups who said “never” (rather than “no, … but”).

has.external.stress has.negative.perception p.no.never
                Yes                     Yes        40%
                Yes                      No      32.4%
                 No                     Yes      55.6%
                 No                      No       5.7%

Logistic regression

We fit two logistic regressions and compared them with a likelihood ratio test. As a null model, we fit the simple logistic regression of no.never as a function of has.negative perception, ignorant of has.external.stress.

glm(formula = no.never ~ has.negative.perception, family = "binomial", 
    data = o)

Deviance Residuals: 
   Min      1Q  Median      3Q     Max  
-1.047  -0.774  -0.774   1.314   1.644  

                        Estimate Std. Error z value Pr(>|z|)    
(Intercept)               -1.053      0.191   -5.51  3.5e-08 ***
has.negative.perception    0.737      0.317    2.33     0.02 *  
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 256.03  on 206  degrees of freedom
Residual deviance: 250.67  on 205  degrees of freedom
AIC: 254.7

Number of Fisher Scoring iterations: 4

The has.negative.perception coefficient is significantly different from zero, indicating that negative perception is associated with participation.

Then we fit a logistic regression for the full relationship described above, which adds the has.external.stress term.

glm(formula = no.never ~ has.external.stress * has.negative.perception, 
    family = "binomial", data = o)

Deviance Residuals: 
   Min      1Q  Median      3Q     Max  
-1.274  -0.885  -0.885   1.354   2.393  

                                            Estimate Std. Error z value Pr(>|z|)    
(Intercept)                                   -2.803      0.728   -3.85  0.00012 ***
has.external.stress                            2.068      0.757    2.73  0.00627 ** 
has.negative.perception                        3.027      0.990    3.06  0.00224 ** 
has.external.stress:has.negative.perception   -2.697      1.048   -2.57  0.01007 *  
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

(Dispersion parameter for binomial family taken to be 1)

    Null deviance: 256.03  on 206  degrees of freedom
Residual deviance: 237.79  on 203  degrees of freedom
AIC: 245.8

Number of Fisher Scoring iterations: 5

All of the non-intercept coefficients significantly different from zero, indicating that negative perception and external stress, separately, are both associated with participation and that the associations are different when combined. Specifically, someone with just negative perception or external stress has higher odds of saying “never” than someone with neither, but the odds are somewhere in between for someone who has both negative perception and external stress.

Finally, we compared the two with a likelihood ratio test.

Likelihood ratio test for MLE method 
Chi-squared 2 d.f. =  12.88 , P value =  0.001594 

The likelihood ratio test finds that the data are significantly more likely given the full model than the null model, suggesting, again, that external stress significantly mediates the relationship between negative perception and participation in activism.

Ordinary least squares

We repeated this analysis with ordinary least squares (OLS) regressions, swapping the likelihood ratio test for an F test. Logistic regression is more appropriate for these data because they have a binary response, but the OLS results are similar and may be easier for some people to understand.

lm(formula = no.never ~ has.negative.perception, data = o)

   Min     1Q Median     3Q    Max 
-0.422 -0.259 -0.259  0.578  0.741 

                        Estimate Std. Error t value Pr(>|t|)    
(Intercept)               0.2587     0.0383    6.75  1.5e-10 ***
has.negative.perception   0.1631     0.0689    2.37    0.019 *  
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

Residual standard error: 0.458 on 205 degrees of freedom
Multiple R-squared: 0.0266, Adjusted R-squared: 0.0219 
F-statistic:  5.6 on 1 and 205 DF,  p-value: 0.0188 

lm(formula = no.never ~ has.external.stress * has.negative.perception, 
    data = o)

   Min     1Q Median     3Q    Max 
-0.556 -0.324 -0.324  0.600  0.943 

                                            Estimate Std. Error t value Pr(>|t|)   
(Intercept)                                   0.0571     0.0759    0.75   0.4526   
has.external.stress                           0.2669     0.0874    3.06   0.0026 **
has.negative.perception                       0.4984     0.1679    2.97   0.0034 **
has.external.stress:has.negative.perception  -0.4225     0.1836   -2.30   0.0224 * 
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

Residual standard error: 0.449 on 203 degrees of freedom
Multiple R-squared: 0.0734, Adjusted R-squared: 0.0598 
F-statistic: 5.36 on 3 and 203 DF,  p-value: 0.00142 
Analysis of Variance Table

Model 1: no.never ~ has.negative.perception
Model 2: no.never ~ has.external.stress * has.negative.perception
  Res.Df RSS Df Sum of Sq    F Pr(>F)   
1    205  43                            
2    203  41  2      2.07 5.13 0.0067 **
Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1 

Not experimental

We intuitively suspect a particular direction of causality and have fit models that suggest that direction, but the questionnaire gives us no formal idea about the direction of causality; for example, it could be that the choice never to participate leads to overstatement of reasons against participation and of external stresses.


External stresses mediate an association between negative perception of activism and a decision never to participate in Occupy CUNY. For interpreting the direction of this relationship, the following plot is helpful.

plot of chunk summary.plot

Negative perception makes people more likely to say they will “never” participate, regardless of the level of external stress, and having external stress makes this difference larger.

Ideas for future study

It would be nice to pin down our nebulous concepts of external stress and negative perception. Rather than qualifying the sorts of stresses and perceptions, we simply ignored which boxes people checked for questions 4c and 7c. It may be informative to see how the particular box that was checked relates to the question 8 result or other results.

We could also test the same relationship with the larger and more general Occupy Research General Survey. It contains questions similar to the ones from the OccuPAR questionnaire that we used as indicators of external stress, perception of activism and participation in activism, but it has more responses from a wider movement.


The conclusion of this study can be applied to the marketing surrounding activism. Aside from trying to influence people’s perception of the value of activism, marketing should consider activists’ external stresses.

People who are trying to recruit and sustain activists should focus on people with fewer external stresses or should provide support to current and potential activists who have more external stresses.

People who are trying to suppress activism should focus on people with more external stresses or should try to make the lives of potential and current activists more stressful.

The research team

The research team was initially composed of twelve individuals, including four CUNY undergraduate students, seven CUNY graduate students, and one alumnus. Most individuals on the research team had previous experience as student activists addressing issues at CUNY schools. The group met consistently from February, 2012 to May, 2012 in order to develop a survey designed to find the supports and barriers to student activism on CUNY campuses. During this time, one undergraduate student and one graduate student dropped out of the study. The research team had ten members when the survey was finalized: Sarah Zeller-Berkman, Keiko Matsuura, Alexis Halkovic, Jen Tang, Patrick Sweeney, Do Lee, Caro Muñoz-Proto, Akemi Nishida, Farhana Miah, and Mimiko Watanabe. Tom joined this team at the OccupyData hacakathon on March 1 to assist in the present analysis.

Source code and other documentation are here, as is the questionnaire.