This is a field trial with a randomized pre-test post-test control group design. This trial is the second phase of an exploratory mixed methods research. Prior to this trial, a qualitative study through in-depth interviews to tour guides from 11 language divisions and also policy makers of Indonesian Tour Guide Association Bali branch (HPI Bali). The education model for tour guides were developed based on the integration of theory of planned behavior (TPB) and identity theory. This educational model is expected to improve the behavior of tour guides in providing travel health information to tourists they serve, which is currently still lacking. However, due to the COVID-19 pandemic, the measurement of behavior can not be conducted. Therefore, this trial aims to test the efficacy of the comprehensive education model to improve the indirect and direct determinants of the behavior namely attitude, subjective norms, perceived behavioral control, role identity, actual behavioral control and behavioral intention.

The research hypothesis are:

1. The comprehensive travel health education model improves the attitude of tour guides towards providing travel health information to tourists?
2. The comprehensive travel health education model improves the subjective norms of tour guides towards providing travel health information to tourists?
3. The comprehensive travel health education model improves the perceived behavioral control of tour guides in providing travel health information to tourists?
4. The comprehensive travel health education model improves the role identity of tour guides as a travel health promoter for tourists?
5. The comprehensive travel health education model improves the actual behavioral control of tour guides in providing travel health information to tourists?
6. The comprehensive travel health education model improves the intention of tour guides to provide travel health information to tourists?

1. Study design: This is a field trial with a randomized pre-test post-test control group design. This trial is the second phase of an exploratory mixed methods research.

2. Study setting: Data collection will be conducted in Bali, Indonesia, started from September 2021 to February 2022.

3. Study population: The target population in this study are all general tour guides in Bali, Indonesia. The accessible population are general tour guides who are registered as members of HPI Bali who can be reached during the data collection period.

4. Sample size: The calculation of the minimum sample size in this study was carried out with the assistance of the WHO Sample Size Determination software using the hypothesis test formula for the mean difference of two populations (one-sided).

   Based on the sample size calculation with the above formula, the minimum sample size for each group is 69 tour guides. Taking into account the drop-out rate of 10%, the number of samples for each group is 76 tour guides. Thus, the total number of samples for intervention and comparison groups in this study is 152 tour guides.

5. Sample selection: The sample selection was carried out with disproportionate stratified random sampling using a sampling frame namely the list of tour guides who are registered as the members of Indonesian Guides Association (HPI) Bali Branch from 11 language divisions. The number of samples in each division for intervention and control group ranged from 5 to 8 subjects.

6. Sample allocation: Permuted block randomization.

7. Study variables:

   1. Dependent variable is comprehensive travel health education model.

   2. Intermediate variables are:

      • attitude towards providing travel health information to tourists,
      • subjective norms towards providing travel health information to tourists,
      • perceived behavioral control of providing travel health information to tourists,
      • role identity,
      • actual behavioral control.

   3. Dependent variable is intention to provide travel health information to tourists.

   4. Controll variables are:

   Controlled by design:

   • history of formal health education,
   • ownership of a smartphone or laptop with WhatsApp and Zoom application,
   • familiarity with Zoom usage,
   • familiarity with WhatsApp usage.

   Controlled by analysis:

   • age,
   • sex,
   • education,
   • length of work,
   • employment status,
   • tourists' country of origin,
   • type of tourism activities,
   • employer's policy.

8. Research instrument:

The instrument that will be used in this study is a questionnaire. The measurement using questionnaire will be conducted before the intervention (pre-test) and after the intervention (post-test). The developed questionnaire consists of several parts, namely the identity or socio-demographic characteristics of the subjects, job characteristics, characteristics of tourists served, attitudes, subjective norms, perceived behavioral control, role identity, actual behavior control and behavioral intentions.

The questionnaire was developed based on the guidelines for developing a questionnaire for TPB-based interventions and the results of the qualitative study. The assessment for each statement item on the attitude, subjective norm, perceived behavioral control and role identity, uses a Likert scale consisting of seven scales, namely: 1 = strongly disagree, 2 = disagree; 3 = disagree, 4 = undecided, 5 = somewhat agree, 6 = agree; 7 = strongly agree.

The questionnaire on the actual behavior control section consists of questions related to knowledge about prevention and first aid for tourism health problems. Questions to measure knowledge were developed based on the material provided during the intervention. The correct answer for each knowledge question will get a score 1, while the wrong or 'do not know' answer will be given a score 0.

The questionnaire on the behavioral intention section is also written in statement format. The assessment for each statement item uses a Likert scale consisting of seven scales, namely: a scale of 1 = strongly not intending, 2 = not intending; 3 = lack of intention, 4 = undecided, 5 = somewhat intend, 6 = intend; 7 = very intent.

The validity and reliability of the questionnaire will be tested before it being used in data collection. Twenty tour guides will be involved in the pilot test of the questionnaire. If necessary, improvements will be made to the questionnaire based on the test results. The questionnaire is self-administered and will be given to the subjects in digital format.

9. Data analysis:

1. Descriptive statistic Quantitative data obtained from the questionnaire will be analyzed descriptively by calculating the frequency distribution for data on a categorical scale and by calculating the total value, mean, median, standard deviation and range of values from minimum to maximum on continuous scale data. On the variables of attitude, subjective norm, perceived behavioral control, and role identity, a positive statement will get a score of 1 = strongly disagree; 1 = strongly disagree; 2 = disagree; 3 = disagree; 4 = doubtful; 5 = somewhat agree; 6 = agree; 7 = strongly agree. As for negative statements, the score given is the opposite, namely a score of 7 = strongly disagree; 6 = disagree; 5 = disagree; 4 = doubtful; 3 = somewhat agree; 2 = agree; 1 = strongly agree. The actual behavioral control score was 1 for correct answer and 0 for wrong answer and 'do not know' answer.

2. Chi square test Comparison of sample characteristics in the intervention and control groups in terms of control variables (age, gender, education, length of work, employment status, tourists' country of origin, type of tourism activities and employer's policy) was also carried out using the chi square test. Through this test, it will be known whether there are differences in the two groups based on these variables. If the p value > 0.05, the two groups are the same (comparable), whereas if p value ≤ 0.05 then the two groups are different in terms of the variables tested. Variables that have a p value ≤ 0.05 will be tested in multivariate analysis.

3. Normality test Data normality test is conducted to determine whether or not the data on attitude, subjective norms, perceived behavioral control, actual behavior control, role identity and behavioral intentions scores are normally distributed, as a prerequisite for being able to perform parametric statistical tests. Based on the sample size, the normality test will be carried out using the Kolmogorov-Smirnov test. If the normality test produces a p value > 0.05 then the data distribution is normally distributed. On the other hand, the data is not normally distributed if the p ≤ 0.05.

4. Homogeneity test The purpose of the homogeneity test is to find out whether the two groups to be compared have the same variance. According to the sample size, homogeneity test will be carried out using Levene's test. If the results of the homogeneity test show a p value > 0.05 then the variance between groups is homogeneous. On the other hand, if the p value ≤ 0,05, the variance of the two groups is not homogeneous.

5. Mean comparison test The mean comparison test will be conducted to determine whether there is a difference in the mean between the intervention and the control group (difference in the mean of two independent samples). The mean comparison procedure used in this study is the independent samples t test. However, testing with independent samples t test is a parametric test which assumes the data is normally distributed and both groups have the same variance. The interpretation of the test results is that Ho is accepted if the p value > 0.05 or the zero value lies within the confidence interval for the difference in the mean, on the other hand Ho is rejected if the p value ≤ 0,05 or the zero value lies outside the confidence interval for the mean difference of the two groups. If the data is not normally distributed, the data will be transformed. Furthermore, if the transformation cannot make the data normally distributed, the data will be analyzed by non-parametric statistical tests, namely the Mann Whitney U test.

   In addition to the comparison of the mean between the intervention and comparison groups, a comparison test of the means was also carried out for the control variables. For categorical-scale control variables with two categories, a t-test was performed, such as a comparative analysis of the mean for the intervention and control groups. Meanwhile, for control variables that have more than two categories, an analysis of variance (ANOVA) test will be carried out. If the assumptions for the parametric statistical test are not met, the Kruskal-Wallis test is performed.

6. Multivariate analysis After the bivariate analysis, a multivariate analysis was carried out which aims to control the inter-correlation between the variables under study. In multivariate analysis, all dependent variables are tested simultaneously. If the assumptions for parametric statistical analysis are met, the statistical test used is multivariate analysis of variance (MANOVA). MANOVA is chosen because there is more than one continuous-scale dependent variable to be tested and more than one categorical-scale independent variable with two or more categories. If the assumptions are not met, then a non-parametric MANOVA statistical test is performed.

Inclusion Criteria:

1. Active members of HPI Bali,
2. Age less than 55 years old,
3. Have worked as a tour guide for one year or more,
4. Have a smart phone or laptop that equipped with Zoom and WhatsApp application,
5. Familiar with Zoom meeting,
6. Using WhatsApp actively.

Exclusion Criteria:

1. Holding a structural position at HPI Bali,
2. Unwilling to work as a tour guide after the re-opening of tourism,
3. Have a formal health education background,
4. Unwilling to participate in this study.