GAMAAN aimed to measure and document attitudes and opinions in a closed society, which cannot be obtained using conventional methods. To this aim, GAMAAN uses multiple chain-referral sampling method through different online platforms and social media (Telegram, Instagram, WhatsApp, and Twitter).
Studies employing opt-in online surveys face methodical challenges that are inherent to online sampling methods, which are nevertheless becoming the norm. These shortcomings include the so-called network effect, which means that the survey is more likely to reach respondents who hold beliefs similar to those held by the organizers, as well as self-selection, which means that those with a special interest in the survey topic are more likely to participate.
To reduce these effects, GAMAAN’s surveys are spread through various social-media groups, channels, and pages representing radically diverse social layers of society and political perspectives. Using multiple chain-referral sampling to reach a diverse audience, surveys are shared with and by online pages and channels belonging to specific groups, such as minorities’ and pro-regime groups’ networks, as well as a mass audience consuming social, political, and entertainment contents. The targeted Instagram pages and Telegram channels range between 10 and 100 thousand followers, while those with a general audience range up to 1 or several million followers. These measures also increase the sample size, further minimizing bias.
Samples obtained from online surveys generally do not properly overlap with the target population’s characteristics. Weighting is used to obtain a representative sample. This technique balances the sample in accordance with characteristics of the target population in question.
GAMAAN employs a weighting method, raking, and also cell weighting with interlocked targets, to generate a representative sample from the refined sample. As a study conducted by the PEW Research Center shows, raking method is among the most effective and reliable for samples derived from online surveys. As suggested by PEW, a variable reflecting respondents’ political behavior is introduced in surveys and used for weighting. This decreases the sampling bias while increasing generalizability.
To obtain a representative sample, the refined sample drawn from respondents living inside Iran is weighted based on demographic factors of sex, age group, education level, province, urban/rural areas, and respondents’ voting behavior.
GAMAAN uses the official data from the 2016 National Population and Housing Census to extract the target population characteristics.
The results of GAMAAN’s surveys are verified using different reliability checks. One of the methods for examining the reliability and generalizability of a weighted sample is to compare the results from the weighted sample against external evidence. GAMAAN checks the congruence of its final sample on the external factors such as employment rate, household income, language speaking at home, and the results of non-sensitive questions with findings of reputable pollsters.
Furthermore, to estimate the reliability of the results drawn from the weighted sample, computations are run once again using the matching method, and the results were compared against the original results.
All GAMAAN’s survey reports discuss the employed sample balancing, weighting methods, and reliability checks as well as the characteristics of the sample demography and the target population.