Page 124 - SUSTAINABILITY ISSUES & COVID-19

3.3 Independent Variables
Prior research measures Social Media Governance (SMG) concept using different methods, such as interviews and
policy reviews (Haynes, 2016; Johnston, 2015; van den Berg & Verhoeven, 2017). Other quantitative research from Avery
and Graham (2013) use surveys to collect social media applications, the perception of a citizen, and self-satisfaction.
However, this research did not analyze SMG implementation that is embedded in an organization’s strategies. Linke
and Zerfass (2013) examined SMG in a more comprehensive approach because it viewed governance as an overall
concept, not only as policies and guidelines. Linke and Zerfass (2013) divide the framework into four focal areas: social
media activities, social media strategies, social media skills, and regulatory frameworks.

3.3.1 Instrument development
The questionnaire was adopted from Linke and Zerfass (2013) and translated in Indonesian. Four additional questions
adopted from Zammuto and Krakower (1991) were inserted in the survey to measure organizational culture as a
moderator variable. Two rounds of pilots were conducted to refine the instruments. First, this survey was examined by
three people to check whether the translation is accurate, and the question is straightforward. Second, surveys were
distributed to 100 random people using a premium online survey. These 100 sample respondents are excluded from
real analysis. The results were examined to ensure the reliability and validity of the instruments.

3.3.2 Response rate and sample selection
The survey was conducted from December 2019 to February 2020. Questionnaires were distributed to 400 respondents
from 40 organizations identified by purposive sampling, and 50 respondents chosen using convenience sampling. From
a total of 450 surveys distributed, only 205 surveys returned. From all returned questionnaires, only 171 questionnaires
meet the criteria and represent 51 organizations. Thus, the response rate is 38%. Similar research from Ralston, Terpstra‐
Tong, Terpstra, Wang, and Egri (2006) has a 44% response rate.

3.3.3 Instrument Reliability and Validity
Reliability is measured using Cronbach’s Alphas, which should be 0.7 or more (Straub, 1989). Validity is measured by
Kaiser Mayer Olkin (KMO), loading factors, and cross-loading. The acceptable score is 0.5 for KMO, loading factors, and
cross-loading convergent validity (Yamin & Kurniawan, 2011). As reported in Table 1, the instruments are reliable and
valid because Cronbach’s Alphas are more than 0.7. Additionally, KMO scores, loading factors, and convergent validity
are more than 0.5. Discriminant validity also shows satisfying results because the correlation between indicators from
the same construct is higher than the correlation between indicators from different constructs.

Table 1 Test Result of Validity and Reliability
Cross Loading
Convergent
Variable Cronbach Alpha Kaiser’s MSA Factor Loading Same Differenbt
Validity
Constructs Constructs
Social Media 0,2216;
Activities -0.0197;
1 - 1 1 1
0.2941
0.4877
Social Media -0.3896;
Strategies 0.8236 0.695 0.6392- 0.567 0.7531 0.2941
0.3423
0.4877;
Social Media 0.2216
Skills 0.9167 0.914 0.6588- 0.600 0.7748 -0.2917
0.8601
-0.3896;
Regulatory 0.8108- -0.3896;
Framework 0.9398 0.931 0.714 0.8447 -0.2917;
0.8974
-0.0197

Source: Adapted from STATA

International Conference on Sustainability 123
(5 Sustainability Practitioner Conference)
Th